INTERIM REPORT I PAYMENT CARDS

Transcription

1 EUROPEAN COMMISSION Competition DG Services Financial Services (Banking and Insurance) INTERIM REPORT I PAYMENT CARDS Sector Inquiry under Article 17 Regulation 1/2003 on retail banking 12 APRIL 2006 Commission européenne, B-1049 Bruxelles / Europese Commissie, B-1049 Brussel - Belgium. Telephone: (32-2)

2 TABLE OF CONTENTS Executive Summary Glossary Section A Industry Background and Outline of the Inquiry I. Organisation of POS Card Payment Systems 1 1. Players in a POS card payment system 1 2. Operation of a POS card payment system 3 3. Fees paid in a POS card payment system 4 4. Conclusion and analysis 4 II. Economics of the Payment Cards Industry 6 1. Survey of seminal economic literature on two-sided industries 6 2. What makes payment card systems different? 7 3. Economic modelling of payment card systems a survey 7 4. Effects of interchange fees 8 5. Competition between payment card systems 9 6. Competition among means of payment 10 III. Data and Methodology Data issues Methodology Sampling Technique Main drawbacks of the technique Data patterns 16 Section B Financial Aspects of the Industry IV. Interchange Fees Purpose of interchange fees Academic explanations Industry explanations Setting of interchange fees International schemes Domestic schemes Level of interchange fees Level of interchange fees: cross-border interchange fees 19

3 3.1.1 Fee structure Comparison of nominal fees (MC/Visa, 2005) Comparison of weighted average fees (MC/Visa) Level of interchange fees: domestic interchange fees for credit cards Fee structure Comparison of nominal fees for credit card transactions (MC/Visa) Comparison of weighted average fees for credit card transactions (MC/Visa) Level of interchange fees: domestic interchange fees for debit cards Not all domestic systems have interchange fees Comparison of weighted average fees (MC/Visa) Comparison of fee levels in domestic card networks (debit and credit cards) Preferential interchange fees in domestic card networks Conclusion and analysis 31 V. Merchant Charges Structure of merchant service charges Description of merchant service charges Merchant service charges for different categories of merchants Merchant service charges: small vs large merchants Merchant service charges: various merchant sectors Blended merchant service charges Countries with no merchant service charges or highly regulated merchant service charges Levels of merchant service charges Credit cards (MC/Visa) EU-25 comparison Country analysis Debit cards EU-25 comparison Country analysis Conclusion and analysis 50 VI. Cardholder Fees Introduction Credit cards Fee per card Correlation between fee per card and interchange fee Card issuance fee Account statement and billing information fee Fee per transaction 58

4 3. Debit cards Fee per card Card issuance fee Account statement and billing information fee Fee per transaction Conclusion 61 VII. Profitability Introduction Profitability trends Credit cards Acquiring business Issuing business Taken issuing and acquiring businesses together Profitability vs interchange fee Debit cards Acquiring business Issuing business Taken issuing and acquiring businesses together Conclusion and analysis 76 Section C Organization of the Industry VIII. Concentration of Acquiring and Issuing Businesses Acquiring Acquiring in international payment networks: credit cards Acquiring in international payment networks: debit cards Acquiring in national payment networks Analysis of the correlation between the HHI level, the number of acquirers and the MSC level Issuing Conclusions and analysis 86 IX. Integration of Card Payment Systems Different degrees of vertical integration of card payment systems in the EU Joint ventures for acquiring services Implications of clearing arrangements Membership requirements to buy processing services Analysis Conclusion and analysis 94

5 X. Governance in Card Payment Systems Classes of membership in card payment systems International systems Domestic payment card schemes Reasons for distinguishing between classes of membership Reasons quoted for distinguishing between Principal and Affiliate Members Reasons quoted for distinguishing between different members according to roles played Reasons quoted for the difference between clearing banks and others Reasons quoted for distinguishing members according to ownership status Implications of different classes of membership Collection of business-sensitive data Decision-making on issues affecting intra-system competition Supervisory and sanctioning powers Membership applications Analysis Conclusion 101 XI. Membership Conditions and Fees Selected membership conditions Financial institution requirement Local establishment requirement Joining fees Factors determining joining fees Level of joining fees Double membership in card payment systems Conclusion and Analysis 106 XII. Cross-Border Competition Entry into foreign acquiring markets Central or cross-border acquiring under licence: historical development, current state of play and sectoral bias Country analysis of acquirers involved in cross-border acquiring Analysis of merchants acquired cross-border: country level and general trends Merchants acquired cross-border: national vs multinational merchants Factors boosting central acquiring or cross-border acquiring: country examples Sweden Germany Factors impeding the development of cross-border acquiring Technical standards: communication protocols, security standards and certification 116

6 7.2 Membership requirements and fees Interchange fee arrangements Conclusion and analysis 119 Section D Other Important Characteristics of the Industry XIII. Network Rules Other than on Interchange Fees Co-branding Rules and practices Analysis Surcharging and no discrimination rule Rules and practices Analysis Conclusion and analysis 125 XIV.Non-Price Competition Variables for Credit and Debits Cards Main factors in non-price competition Non-price competition in old vs new Member States Non-price competition for large vs small firms Conclusion and analysis 129 XV. Interest Free Periods and Float in POS Card Transactions Introduction Analysis of free funding period and net float per card brand Analysis of free funding period and net float per Member State EU-25 Overview Average transfer periods in the EU Average free funding period in the EU Average net float in the EU In some EU countries banks appear to treat all card brands like one single card Conclusion and analysis 138 Section E - Summary and Analysis XVI.Summary and Analysis of the Findings Financial analysis Barriers to competition Barriers to competition between banks and non-banks Barriers to competition between different card payment systems Distortion of competition between different means of payment 144

7 Annexes Annex 1 Methodology for calculating weighted average interchange fees Annex 2 Analysis of fee structure for international debit card systems (MC/Visa) Annex 3 Joining fees for open domestic payment systems Annex 4 Structure of joining fees in one card system (Country A) Annex 5 Econometric Study

8 Executive Summary A. Purpose of the Commission s sector inquiries Competitive and efficient financial services markets are vital for the success of the European economy, in serving businesses and consumers efficiently to help deliver strong economic growth and sustain high levels of employment. The drive to deliver an efficient and competitive financial services industry in Europe can therefore make an important contribution to achieving the Lisbon goals. The Financial Services Action Plan (FSAP) has made a significant contribution to developing the framework to support financial services integration in Europe. Following the FSAP, in its recent White Paper Financial services policy , the Commission set out its future strategy. The strategy has five priorities: to dynamically consolidate progress and ensure sound implementation and enforcement of existing rules; to drive through better regulation principles in all policy making; to enhance supervisory convergence; to create more competition between service providers, especially those active in retail markets; and to expand the EU's external influence in globalising capital markets. The Commission s sector inquiries into financial services specifically into retail banking and business insurance are a central part of this post-fsap agenda. The retail banking inquiry will make a significant contribution here, particularly to the second and fourth priorities. The aims of the sector inquiry into retail banking are to: improve the Commission s market knowledge of retail banking, notably to provide an empirical basis for implementing the post-fsap strategy for retail financial services; give all stakeholders concrete information about potential market failures, enabling them to resolve these problems where possible; identify issues that require investigation and possibly remedy under the European competition rules (Articles 81 and 82); and provide a framework for National Competition Authorities (NCAs) and the Commission, to ensure that the many ongoing competition procedures are coherent. In its inquiry into retail banking, the Commission is looking at the markets for core retail banking services, particularly (1) current accounts and related services, and (2) payment cards. This interim report into competition in payment cards is complementary to the work on current accounts and related services. The findings from both aspects of the inquiry will be considered together, allowing the Commission to broaden and deepen its understanding of competition in EU retail banking. The final report on the sector inquiry into retail banking, covering current accounts (and related services) and payment cards, will be published by the end of Legislation has been proposed to create a Single Euro Payment Area (SEPA) in the EU, to make cross-border payments in euros in the EU as easy and affordable as domestic payments. This alone could save the EU economy between 50 and 100 billion per year. 2 1 Available at 2 See "Time to Move Up A Gear" - The European Commission's 2006 Annual Progress Report on Growth and Jobs. Available at: i

9 The European payment cards industry is large and handles a significant part of retail sales in Europe. Total sales volumes with point-of-sale card transactions in the EU in 2005 were more than 1350 billion. It is estimated that businesses in the EU paid more than 25 billion in fees in It is estimated that cards alone account for up to 25% of retail banking profits. However, the payment cards industry shows evidence of continuing fragmentation and the inquiry has found striking differences in the levels of prices and profitability across Member States. The findings of the inquiry into core retail banking, and in particular payment card systems, will provide valuable evidence to shape the future development of the Single Euro Payment Area (SEPA) project. In particular, the inquiry aims to show how differing forms of organisation, structure and governance of payment systems in the EU can produce differing competitive outcomes. The evidence gathered for the inquiry suggests that the characteristics of some payment systems lead to significantly higher prices for firms and consumers in some Member States. As work continues to develop the appropriate principles and structures to support SEPA and its Payment Cards Framework (PCF), significant consideration should be given to the findings of the Commission s retail banking inquiry. Format of the interim report The interim report into payment cards is structured as follows: Section A sets out the context of the inquiry and its methodology, outlines some structural and product features of the industry and describes relevant economic theories; Section B examines financial aspects of the payment cards industry, particularly revenue sources and overall profitability; Section C examines the structure and governance of the industry and highlights potential barriers to competition; Section D examines non-price factors for competition in card payments; and Section E summarises the report s findings. This interim report is a summary of the Commission s findings on competition in payment cards. The analysis is based on an extensive market survey conducted by the Commission throughout the second half of Thus, the Commission s inquiry has been able to draw on a rich and detailed evidence base. It is hoped that this evidence base will enable the Commission, together with market participants and authorities, to reach clear and robust conclusions about competition in payment cards in the EU. However, this interim report is only the first stage in the process. The Commission is keen to engage in a dialogue with market participants and authorities about the report s findings and appropriate ways forward. Therefore, the Commission highlights five sets of issues for consultation and welcomes the views and perspectives of all stakeholders on this interim report. 3 Estimate derived from Payment Cards Report, RBR, 2005, London. ii

10 B. Interim findings on competition in payment cards Card payment systems enable consumers to use and businesses (merchants) to accept plastic cards as a method of payment. When a cardholder uses a card to purchase a good or service, the bank that issued the card, the issuer, debits the retail price from the cardholder s account. The issuer then pays the bank that acts for the merchant, the acquirer, the retail price less an interchange fee. Finally, the acquirer pays the merchant the retail price less a merchant fee. Retail price plus account charges Card Issuer Retail price less interchange fee Merchant acquirer Retail price less merchant fee Customer/ Cardholder Goods or service Retailer Overall findings The preliminary results of the inquiry show a picture of market fragmentation. While consumers clearly reap benefits from card payment networks in the EU, businesses do less so and largely foot the bill, particularly in the case of credit cards. Some but not all networks offer consumers a means for easy and convenient cross-border payments. This is clearly positive. But businesses, in particular small firms, largely do not benefit from market integration in card payments. This means that in a sector which is key for the European economy, the retail sector, potentially great opportunities for more economic efficiency are foregone for the time being. The inquiry suggests that building a true Single Euro Payments Area (SEPA), one that offers tangible benefits to business and consumers and contributes significantly to growth and competitiveness in this sector, still requires considerable work to be done. It must be recognised that fragmentation in payment markets, and in card payment markets in particular, is partly the result of historical evolution. Fragmentation is due to the way in which payment systems were created and built up in the EU Member States through coordination and cooperation between banks at national level. While bearing this in mind, however, change appears necessary to move forward towards a SEPA. Lessons for building SEPA could also be learnt by looking at efficient payment services provided by existing domestic card networks. Such networks offer in some respects a good value proposition to customers and often charge lower fees to cardholders and business than the large international networks (MasterCard and Visa). A challenge remains in creating efficient cross-border functionalities as a pre-requisite for SEPA. The future choice of SEPA payment schemes by banks is a key issue for the further debate on SEPA. iii

11 The interim findings of the payment cards inquiry are divided into two parts: financial analysis of the industry; and potential barriers to competition Financial analysis of the industry Profitability On the issue of profitability the main findings of the inquiry are: Profitability in card issuing is high and has been sustained over time. The credit card business is particularly profitable, with a weighted average profit-to-cost ratio of 65% for issuing. The average profit-to-cost ratio for debit card issuing is also high at 47%. High profitability is often correlated with high fees charged to merchants and cardholders. The evidence also suggests that even in the absence of an interchange fee, other revenues alone would in many cases generate a healthy profit for issuers. Profitability is higher for credit cards than for debit cards. In both issuing and acquiring activity, credit cards are more profitable than debit cards. Even without interchange fees, card issuing remains profitable. The evidence suggests that card issuing would generate positive profits in 20 out of 25 countries even without interchange fee income. Profitability in card acquiring varies, though is quite satisfactory overall. Credit card acquirers across the EU have a 15% profit-to-cost ratio as a weighted average, while debit card acquirers average around 5%. Profitability is far higher for card issuers than for acquirers. For both debit and credit cards, issuing is significantly more profitable than acquiring. Although this general finding was to be anticipated, the difference in relative profitability is striking. A range of explanations are possible, including the supposition that card issuers may have market power relative to acquirers. Acquiring banks revenues: fees paid by businesses The merchant fee is the price per transaction that a business (or merchant ) pays to the acquirer for accepting cards as a method of payment. The results of the inquiry show that merchant fees vary considerably across the European Union. These differences remain significant when several factors that may affect merchant fees (such as different risk levels) are controlled for. This may indicate that the market for card payment services is not working effectively in many Member States, to the detriment of businesses and consumers. There is evidence of price dispersion at five levels: Businesses in some countries pay a far higher merchant fee on average than others. This pattern is particularly pronounced. Merchants in Hungary, the Czech Republic and Portugal have to pay an average fee of between 2.5% and 3.1% of the total transaction value to accept a MasterCard/Visa credit card. This is 3 to 4 times higher than in Sweden, Finland and Italy. iv

12 Businesses pay a far higher merchant fee on average to accept credit than debit cards. For example, a merchant in the UK pays almost five times as high a fee on average for accepting a MasterCard credit card as compared to a MasterCard debit card. Businesses pay a far higher merchant fee on average to accept cards issued in the international networks than cards issued in the domestic networks. Typically, businesses pay 30-40% lower fees on average for domestic debit card usage than for MasterCard (Maestro)/Visa debit. International payment systems make smaller businesses pay more than larger ones. This does not seem justified solely by transaction volumes. Smaller firms typically pay between 60% and 70% higher fees on average for MasterCard and Visa credit and debit card transactions than larger businesses. In domestic card payment systems, however, the price difference between smaller and larger merchants is only 7% on average. Businesses in some sectors pay much higher merchant fees on average than in others: For instance, florists, restaurants and car rental firms pay a merchant fee twice that of fuel companies and wholesale trade firms. The acquiring banks practice of charging businesses the same level of merchant fee for accepting cards issued by different networks is known as blending. Acquirers often apply blending to competing products, such as MasterCard and Visa, in both domestic and international card payment systems. The inquiry has found that the blending of prices may weaken inter-network price competition, which in turn may lead to businesses paying higher acquirer fees. Blending appears to be widespread across the EU25. Issuers revenues: financial transfers from acquiring banks and fees paid by cardholders The interchange fee is the fee that an acquiring bank pays per transaction to the issuing bank. It is used as a mechanism to transfer revenues from the acquiring bank to the card issuing bank. The results of the inquiry show that interchange fees vary considerably across the EU. This may indicate that the market for card payment services is not working effectively in some Member States. The levels of price dispersion are similar to those for the merchant fee: Acquirers in some Member States pay far higher interchange fees on average than in others. This is true for international credit and debit and domestic card transactions. For instance, acquiring banks in Poland pay 8 times as much for a Visa debit card transaction than in the UK. Similarly, the interchange fee for a 50 domestic debit card transaction varies from no fee or less than 10 euro cents in Finland, Denmark, Luxembourg, the Netherlands, Ireland and Belgium to more than 60 euro cents in one particular country. Acquirers pay higher interchange fees on average for international credit and debit card transactions than for domestic debit card transactions. For a 50 euro transaction, for example, an acquiring bank in one EU country would pay on average 39 euro cents on a MasterCard credit card, 27 euro cents on a Visa debit card and zero on a domestic debit card. Many acquirers pay a higher interchange fee on average for domestic MasterCard/Visa transactions than for corresponding cross-border transactions. A transaction is considered to be cross-border when the merchant is located in a different country to the cardholder. In about half of the EU-25, acquiring banks pay considerably more for a domestic MasterCard and/or Visa credit card transaction than for a cross-border one; in some cases up to twice the cross-border fee. In these countries, the MasterCard and/or Visa cross-border rates appear to be used as a minimum benchmark when setting the domestic interchange fee. v

13 Cardholder fees are the fees a cardholder pays to the issuing bank for a payment card. The results of the inquiry show that there is no significant negative relationship between the fee per card and the credit card interchange fee at country and network level. The empirical evidence shows that if the interchange fee increases by 1 Euro only 25 cents are passed on to consumers in lower fees. This result challenges the hypothesis advanced by some industry participants and the economic literature that an increase in interchange fees exactly equals a decrease in cardholder fees. Overall, the inquiry has not confirmed the possible justifications for interchange fees which rely on economic efficiency arguments. Potential barriers to competition The investigation has identified a number of potential barriers to competition in the market for card payment services. These barriers are of a structural, technical or behavioural nature: Structural barriers The vertical integration of card payment systems may impede new entrants, in particular non-banks, from competing with the incumbent in one segment of the market. In some instances vendors of terminals have to compete with an incumbent that not only owns the domestic card payment system but also provides the technical and financial services. Systems in Spain and in Portugal, for example, are highly integrated. In Austria and in the Netherlands, however, the market for processing and acquiring services, respectively, has been opened up after the de-integration of the systems. This has led to lower merchant fees in the Netherlands. Joint ventures between local banks to acquire merchants may remove the competitive pressure on merchant fee levels because merchants only have one offer for the network concerned. Such joint ventures exist for instance in many EU countries for acquiring MasterCard and Visa. Technical barriers Diverging technical standards across the EU may hinder acquirers, processors and terminal vendors from operating efficiently on a pan-european scale. There appears to be significant scope for efficient convergence of technical standards in the payment cards industry. Behavioural barriers Agreement on preferential interchange fees between local banks and high fallback fees for foreign banks may raise the costs for foreign banks wishing to enter the market. This seems to be the situation in at least Portugal and Austria. Bilateral clearing arrangements between local banks could make market entry more difficult. New entrants depend on sponsoring banks, who have little incentive to sponsor potential competitors. This seems to have been the situation in the UK. Similar structures are found in Ireland and Finland. vi

14 Some governance arrangements within card payment systems risk distorting conditions for competition between members, in particular between new entrants and the incumbent banks. For instance, in some networks associate members have to communicate business-sensitive information to the principal members without reciprocal information sharing. In other systems, decision-making on issues affecting intra-system competition, such as fees, membership rules and technical specifications, is reserved to the principal members. Some payment system membership requirements may hinder non-banks from domestic acquiring and new entrants from cross-border acquiring. Rules which may constitute barriers include requirements to be a financial institution and to have a local establishment. About half of the domestic card payment systems in the EU require issuers and acquirers to be financial institutions. Some systems also require banks to establish a physical presence. In other systems, however, processors may act as acquirers in the domestic debit card system. Similarly, other systems do not require banks to have a local presence to join them. High joining fees for card payment systems and their structure may discourage new entry and expanded card issuing. The high variation of joining fees across the EU for similar card payment systems may also indicate that the level of fees is not objectively justified. For instance, the joining fee varies from zero in some systems to a fee of over 6 million in one country. Joining fees are particularly high in some, but not all, small countries, so the size of the country by itself does not seem to justify the high level. Other network rules may also prevent or make entry more difficult. For instance, the prohibition on cooperative agreements with competing networks or non-banks, cobranding, may hinder domestic debit card payment systems from entering into competition with MasterCard and Visa or retailers or other operators from entering into competition with the incumbent card issuer. Similarly, the prohibition on merchants charging customers for paying by card, surcharging, may hinder the development of alternative non-cash payment instruments, as the true costs are hidden from consumers via cross-subsidisation. C. Possible remedies On the basis of the interim findings, the table below sets out possible remedies. The remedies may be advocacy, antitrust measures and/or regulation. Issue Possible Details remedy High cardholder fees Advocacy Making information on price differentials transparent could help strengthen the demand side. High merchant fees Advocacy Making information on price differentials transparent could significantly strengthen the demand side. High interchange fees (1) Advocacy (2) Antitrust (3) Regulatory (1) Making information on fee differentials transparent may create some limited pressure on networks to lower fees. (2) But effective remedies might require appropriate antitrust or regulatory actions. On us interchange fees Antitrust Could be examined under competition rules. vii

15 Vertical integration Joint ventures (JVs) between acquirers Financial institution requirements Local establishment requirements Excessive joining fees Prohibition of cobranding with nonbanks Bilateral clearing arrangements Governance issues Technical barriers (standards) Blending of merchant fees Prohibition of cobranding Imperfect price signals on payment instruments (1) Advocacy (2) Regulatory (1) Advocacy (2) Antitrust (3) Regulatory Regulatory Regulatory (1) Antitrust (2) Regulatory Antitrust Advocacy (1) Self- Regulatory (2) Regulatory (1) Selfregulatory (2) Regulatory Advocacy Antitrust Regulatory A differentiated and careful approach is needed to remove distortions but preserve efficiencies. Advocacy and discussion with networks should be the preferred approach to address existing situations. For a future SEPA, separation of scheme ownership and other activities could be considered. The case for separating such JVs could be considered. Is addressed in the newly proposed payments directive. Issue for SEPA. Could be removed. Issue for SEPA. Could be examined under competition rules or could be an issue for regulation. Could be examined under competition rules. Creation of multi-lateral clearing facilities is difficult to obtain through competition or regulatory remedies. Could be addressed through regulation or selfregulation by setting some basic requirements for scheme governance and member/stakeholder participation. It may be worthwhile giving self-regulation bodies some limited time to set interoperable standards, but regulation should be considered if this approach does not work. Basic requirement for SEPA. No apparent antitrust remedy. Making information on blending practices transparent could strengthen the demand side. Could be examined under competition rules. Need to explore how to incentivise banks to introduce transaction pricing that leads consumers to choose the most efficient payment instrument. Issue for SEPA. D. Issues for consultation The Commission is keen to engage in dialogue with market participants and authorities about the report s findings and appropriate ways forward. Therefore, the Commission has singled out five sets of issues for consultation, and welcomes the views and perspectives of all stakeholders on this interim report. This will enable all stakeholders to develop a consensus on the key findings of the report and on possible barriers to competition. In the course of this discussion, all parties can then agree on the appropriate way forward. viii

16 The Commission has identified five sets of issues for consultation on its payment cards report: financial analysis of the industry market structures, governance and behaviour future market developments potential solutions to market barriers lessons for SEPA Financial analysis of the industry 1. Are high merchant fees a competitiveness issue for the EU economy? 2. Are there compelling justifications for the comparatively high level of merchant fees observed in some parts of the EU25? 3. In view of the apparent profitability of card issuing, is there a generally applicable justification for substantial revenue transfers through interchange fees in card payment systems? 4. Are the high profits observed due to innovation or do they arise from some kind of market power in a two-sided industry? 5. What pricing practices, rules and legal provisions distort price signals to consumers and the choice of the most efficient payment instrument? 6. Would cost-based pricing promote the use of efficient payment instruments and how could such pricing be implemented? 7. Do currently existing pricing practices have a substantial negative effect on cross-border card usage by consumers? Market structures, governance and behaviour 8. What market structures work well in payment cards? 9. What market structures do not appear to work well / deliver efficient outcomes? 10. What governance arrangements can facilitate competition within and between card payment systems? 11. What governance arrangements can incentivise card payment schemes to respond to the needs and demands of users (consumers and merchants)? 12. What governance arrangements can allow minority participants or minority members to receive appropriate information and participate appropriately in decision-making? 13. What access conditions and fees are indispensable? 14. To what extent is separation between scheme, infrastructures and financial activities desirable to facilitate competition and efficiency? ix

17 Future market developments 15. Are significant structural changes to be anticipated in the payment cards industry? 16. What are the anticipated impacts on the industry of innovation and technological change? Potential solutions to market barriers 17. How can structural barriers to competition, which may arise for instance from the integration of different functions within a payment system or from joint acquiring ventures, be tackled? 18. Are there compelling justifications for the identified possible behavioural barriers to competition? 19. How much need and scope is there for harmonising technical standards in the payment cards industry? How large are the potential benefits and costs of harmonisation? Lessons for SEPA 20. What lessons (best practice) for the design of SEPA schemes can be learnt from existing national and international payment systems? 21. How could competition between schemes in SEPA be strengthened? 22. Which structural and behavioural barriers to effective competition between banks and payment service providers should be removed to achieve SEPA? 23. What governance requirements should SEPA schemes meet? 24. By what means can interoperable communication protocols, security and other technical standards be achieved and certification procedures be limited to the minimum necessary? 25. Do the removal of barriers to competition, the observance of pro-competitive governance and the creation of interoperable standards require (further) regulation? Procedure for consultation on the interim report The consultation on the interim report opened on 12 April Comments on the consultation are requested by 21 June Comments received after this deadline will not be considered. Format for replies: The Commission asks all respondents to the consultation to provide their response by . Comments received may be published on the Commission s website. Respondents should therefore take care not to include confidential information they do not wish to be disclosed. All responses received during the consultation will be studied by the Commission and taken into account in developing its further analysis and findings. The final report of the sector inquiry into retail banking, covering current accounts (and related services) and payment cards, will be published by the end of x

18 Glossary Automated teller machine (ATM): point where consumers can use plastic cards for withdrawing money. Cardholder: the holder of the card, who uses it as a payment instrument. Card acquirer (or acquiring institution): credit institution or other undertaking, and member of a card scheme that has a contractual relation with a merchant. Card brand: the logo of a particular payment card that has been licensed for use in a given territory. Cardholder fee: the one-off or recurrent fee (or a set of fees) paid by a typical cardholder for the ownership and/or use of a classic/standard debit and/or credit payment card (where no special conditions apply), as well for other ancillary services (e.g. account statement information). Card issuer (or issuing institution): credit institution, and member of a card scheme, that has a contractual relation with a cardholder for the provision and use of a card of that card scheme. In a closed system, the card issuer is the scheme owner, while in open systems several credit institutions act as card issuers. Card scheme owner: defines standards, rules, specifications and access policies and governs the card scheme. Four-party system (or open card payment system): the stakeholders involved are 1) the issuer, 2) the acquirer (may be the same as or different from the issuer), 3) the cardholder and 4) the merchant (in the case of ATM transactions it is usually the acquirer that offers its services via the ATM). Simply put, it can be said that the parties involved are the cardholder, the merchant and their banks. Examples are Visa, MasterCard, and several national schemes. Interchange fee: fee paid by an acquiring institution to an issuing institution for each payment card transaction at the point of sale of a merchant. In certain networks, this may be positive in others it is zero. International card system: has an international presence (issuers and acquirers operating in several countries). The fact that the cards issued in one country can be used in another country makes these systems international. Examples are Visa, MasterCard, American Express and Diners. Merchant: the entity that accepts payments by means of cards. Merchant service charge (MSC) (or merchant fee or merchant discount rate): fee paid for each transaction by a merchant to an acquirer, who processes the merchant s transaction through the network and obtains the funds from the cardholder s bank (issuing institution). The transaction is considered to be executed when the corresponding funds, equal to the price of the sold item, are debited from the consumer s account and, after deducting the merchant service charge, are credited to the merchant s account. National payment card system (or national/domestic payment card network/scheme): usually operates within a single country; i.e. the issuer and the acquirer are within the same country. On-us transactions (as opposed to off-us transactions): in a narrow sense, on-us transactions are payment card transactions where the issuing bank and the acquiring bank are identical. This situation is prevalent in closed payment card systems. In a wider sense on-us transactions occur where the issuing bank and the acquiring bank are separate entities but

19 pertain to a common group of banks. This situation typically arises where issuing banks set up a joint venture which acquires merchants. Transactions between this acquirer and its shareholders are often labelled "on-us" transactions, although strictly speaking issuing and acquiring banks are separate entities. Payment card: card that allows the cardholder to make payments for goods and services at POS (point of sale) terminals or remotely (mail order, telephone order, internet) card-notpresent transactions, respectively. It may be one of the following: Debit card: a card that allows the cardholder to charge purchases directly and individually to a current account at a deposit-taking institution (serves as an access device to funds stored in bank accounts). It is recognized that debit cards may also be closely linked to other products offered by banks. Credit card: a card that allows the cardholder to make purchases up to a certain credit amount, which can then be settled in full by the end of a specified period or only in part, with the remaining balance taken as extended credit and being charged interest; credit cards may be linked to a current account at a deposit-taking bank, but also may be linked to an account that has been set up specifically for the use of the credit card. In this report deferred debit card, which is defined as card that allows the cardholder to make purchases but does not offer extended credit (the full amount of the debt incurred has to be settled by end of a specified period), is treated as a credit card. Payment card system (or payment card scheme or payment card network): technical and commercial infrastructure set up to serve one or more particular card brands and which provides the organisation, framework and rules necessary for the brand to function. Point of sale (POS): point where consumers can use plastic cards for payment transactions at a merchant outlet (often a payment terminal). Three-party system (or closed card payment system): the stakeholders involved are: 1) the card issuer and acquirer (it is the card scheme itself that fulfils both functions), 2) the cardholder and 3) the merchant. Examples are Diners, American Express and some national schemes.

20 Section A Industry Background and Outline of the Inquiry

21 I - Organisation of POS Card Payment Systems This chapter sets out the general organisation of POS card payment systems (physical POS transactions). We explain the operation of a typical POS payment card transaction and analyse which players provide what kind of services in order for such a transaction to be completed. These technical explanations will provide the basis for our further analysis of the economics of the card payments industry and of the vertical integration and the governance of POS card payment systems. We also look at the pricing of services in a POS card payment system. POS card payment systems enable consumers to use plastic cards for payment transactions at the point of sale (POS), which is most often a payment terminal in a merchant outlet. These POS systems are to be distinguished from ATM card payment systems, which enable consumers to use plastic cards for withdrawing money from automated teller machines ( ATM or cash machines ). In practice, POS and ATM systems may be combined so that consumers can use one plastic card both for payment at POS terminals and for withdrawing cash at ATMs, as well as for other ATM services such as printing of statements, balance reporting, credit transfers, etc. This sector inquiry concentrated on POS card payment systems and all subsequent observations therefore cover POS systems only. Where observations exceptionally relate to ATM systems as well, this will be spelled out explicitly. 1. Players in a POS card payment system POS card payment systems involve a wide range of services and service providers. The graph below gives a structural overview of a de-integrated POS system where the roles of scheme ownership, network operation and financial services are attributed to different entities. Graph 1 The above graph shows three main groups of players: (i) cardholders and merchants (ii) scheme owner (iii) issuers and acquirers. Cardholders and merchants engage in a payment transaction through the intermediary of banks and scheme owners. The cardholder receives payment services and credit services from the entity that issued the card (the issuer). The merchant receives payment services from the entity that deals with the merchant (the acquirer). Acquirers may also be issuers. When the cardholder uses the card to buy from the merchant, the merchant receives from the acquirer the retail price less a merchant service charge. The issuer pays the acquirer the 1

22 retail price minus or plus any interchange fee 4. This interchange fee is determined by the card association members of, for instance, MasterCard and Visa. As well as the interchange fee from the acquirer, the issuer receives from the customer the payment, any annual fee, any interest payment on debt outstanding, late payment fees, etc., and might conversely give the customer rebates, loyalty rewards and the like. Issuers issue cards to cardholders and acquirers recruit merchants for payment card acceptance. Payment card issuing is the business of distributing payment cards to consumers on own account and risk while payment card acquiring is the business of contracting merchants for payment card acceptance on own account and risk. Both activities involve certain financial risks with regard to the settlement of a payment card transaction. An acquirer in particular risks losing money on chargeback claims of cardholders. An acquiring bank may be faced with chargeback claims up to several months after it has credited the merchant. If the merchant goes bankrupt in the meantime and if cardholders claim back their money, the acquiring bank may bear the financial costs of the chargeback claim vis-à-vis the issuing bank. Typically, scheme owners reserve issuing and acquiring to credit institutions or entities controlled by credit institutions. Acquiring typically involves the marketing of card acceptance to merchants and therefore requires sales staff. Acquirers also provide customer service to merchants (e.g.: they defend them against chargeback claims of cardholders, check claims that money has not been transferred, etc.). An issuing processor opens and manages the cardholder s account on behalf of the issuer, books card transactions on these accounts, authorises card transactions on behalf of the issuing bank, sometimes arranges the clearing and settlement, provides cardholder statements and sometimes operates a cardholder call centre (for lost and stolen cards) and sometimes also handles chargeback claims of cardholders. An acquirer processor opens and manages the merchant s account on behalf of the acquirer, forwards authorisation requests to a switch (or switches authorisation requests directly to the issuer or issuing processor) and sometimes also supplies voice authorisation centres, books transactions on merchant accounts, charges merchant discount rates to merchants and produces merchant statements. Many acquirers also rent out POS terminals to merchants. Issuing and acquiring processing is often done by the issuers and acquirers themselves. The scheme owner is responsible for: (i) granting licenses (and membership status) to independent financial institutions for the use of a card logo and for performing issuing and acquiring services within the network; it may also (ii) certify non-financial institutions for performing technical activities such as clearing and processing within the system; it usually (iii) sets the network rules and the technical (message) standards; and it (iv) implements these network rules and standards by executing audits at member banks and certificate holders and by organising arbitration in the case of settlement disputes. The graph shows other players who may or may not participate in a POS payment card system, depending on the degree of vertical integration of that system. In a largely de-integrated system, there is scope for competition between non-bank entities that (i) produce payment terminals, (ii) rent out terminals, including maintenance services, (iii) switch transactions between banks, (iv) process transactions on behalf of the issuing bank and/or the acquiring bank, and (v) produce, personalise and destroy payment cards. In many European POS card payment systems, these technical services are concentrated in the hands of a single network service provider. In the Dutch debit card system PIN, for instance, the switching, processing and clearing is done by the inter-bank association Interpay, while in the Danish Dankort debit card system the inter-bank association PBS processes, clears and settles card transactions and even 4 In POS systems, interchange fees are typically paid by acquirers to issuers, but in principle they could go either way and there are systems where no such fees are charged. 2

23 acquires merchants for Dankort acceptance. In some systems, these technical services are provided by the scheme owner, who then not only sets the rules and standards but is also involved in the operational aspects of the payment card system. 2. Operation of a POS card payment system The operation of a POS payment card system depends naturally on the individual structure of each payment card system. The graph below gives a rough illustration of a typical transaction within a highly de-integrated POS payment card system. Graph 2 The numbered arrows in the graph are explained below: (1) The cardholder uses his payment card at a POS terminal of a merchant. (2) The card transaction goes from the POS terminal via the network of a datacoms provider to... (3) the acquiring processor. (4) The acquiring processor sends an authorisation request to the switch. (5) The switch routes this request to an issuing processor operating on behalf of the issuing bank. (6) The issuing processor authorises the transaction after verifying (online or offline) whether sufficient funds are available on the cardholder s current bank account to execute the transaction or after having checked a so-called black list containing stolen and lost cards. (7) Issuing banks may (sometimes) immediately debit the cardholder s account. (8) The authorisation response (approval/refusal) goes from the issuing bank to the issuing processor. (9) The issuing processor routes the authorisation response to the switch. (10) The switch routes the authorisation response to the acquiring processor. (11) The acquiring processor routes the authorisation response via the datacoms provider (12) to the POS terminal at the merchant outlet. (13) The POS terminal sends an acceptance acknowledgment via the datacoms provider to (14) the acquiring processor. (15) The acquiring processor forwards this message to a clearing house, which sends information on all successful transactions in batched form (a package of several messages) to the acquiring banks for payment to the merchant s account and to the issuing banks for debiting from the cardholder s account. After clearing, a transaction is settled. As clearing and settlement may occur bilaterally between pairs of banks, or multilaterally if a POS system uses a common multilateral settlement platform which credits and debits member banks on specific accounts on the basis of clearing messages. 3

24 3. Fees paid in a POS card payment system What is considered here is essentially the technical execution of a POS transaction. Subsequently, we will analyse the commercial aspects of a POS transaction. The flow of fees during a POS transaction is depicted in the graph below. Graph 3 Cardholder Sells goods at price p Merchant Pays p + f, where f = cardholder fees Pays p-m here m=msc Issuer Acquirer Pays p - a where a = interchange fee Network Operator When a cardholder uses his/her card to buy at a merchant outlet, the merchant receives from the acquirer the retail price less the merchant discount rate or merchant service charge (MSC). The issuer pays the acquirer the retail price minus an interchange fee 5. In addition, the issuer moreover receives annual cardholder fees, interest payments on any debt outstanding, late payment fees, etc., and might conversely give the customer rebates, loyalty rewards and the like. We thus see that there are two consumer groups that may be charged the costs of services provided in a POS system. The question as to what extent from an efficiency viewpoint issuing and acquiring banks should subsidise card usage by charging a zero (or even negative) fee to cardholders, while recouping the corresponding costs from merchants through interchange fees, has been widely debated in the academic literature on payment cards and more generally on what are termed two-sided markets. The main elements of this discussion are summarised in the next chapter. 4 Conclusion and analysis A POS card payment system may involve a large number of service providers and there is wide scope for competition between banks and non-bank entities within such a system. Issuing banks provide services to cardholders and acquiring banks provide services to merchants. Other service providers involved in the technical aspects of the cards business are: processors, switches, terminal producers, card producers and telecom providers. These technical 5 In practice, interchange fees go from acquirer to issuer, but in principle they could go either way or be zero. 4

25 services are needed by banks in order to provide their financial services relating to card issuing and/or merchant acquiring. As far as pricing is concerned, banks charge cardholder fees, interest, money exchange fees etc. to cardholders and charge a merchant service charge (also referred to as merchant discount rates or merchant fees) to merchants. In some systems, an acquiring bank may also be required to pay an interchange fee to the issuing bank, which is a fee per card transaction. The scheme owner finally charges membership and licensing fees to both types of banks. 5

26 II. Economics of the Payment Cards Industry 1. Survey of seminal economic literature on two-sided industries Rochet and Tirole 6 (2005) have recently proposed a formal definition of two-sided industries: A market [an industry] is two-sided if the platform can affect the volume of transaction by charging more to one side of the market and reducing the price paid by the other by an equal amount; in other words, the price structure matters and platforms must design it so as to bring both sides on board. To satisfy this definition, the relationship between end-users must be fraught with residual externalities that customers cannot sort out themselves. Payment card systems would be two-sided industries because they have the following characteristics. Firstly, payment card systems would not only serve two distinct groups of customers (cardholders and merchants) but would also have a joint demand, in the sense that they provide a service only if both cardholders and merchants jointly agreed to use a card for a transaction. The demand on one side of the market would thus likely vanish if there were no demand on the other regardless of the price 7. Secondly, payment card systems would have network externalities, in particular what Evans and Schmalensee (2005) call indirect externalities. These indirect network effects would arise because more cardholders in the system make payment card systems more valuable for merchants and more merchants make payment card systems more valuable for cardholders 8. Recent evidence provided by Rysman (2004) shows a regional correlation between consumer usage and merchant acceptance for the four major payment card systems (American Express, Discover, MasterCard and Visa). This correlation suggests the existence of a positive feedback loop between consumer usage and merchant acceptance 9. Besides indirect externalities, Rochet (2003) considers that payment cards would also exhibit usage externalities, which he considers fundamental since they affect the costs and benefits of both parties to the transaction, even though they are not reflected in the fees. For example, when a consumer uses a card, this may benefit the merchant in that the latter does not incur the cost of handling and holding the cash until it can be deposited in a bank. Similarly, when a merchant decides to accept a card for a transaction, this may benefit card users because they do not incur the cost of finding an ATM and withdrawing money (assuming they pay cash). 6 Rochet, J.-C. and J. Tirole, 2005, Two-Sided Markets: an Overview, IDEI Working Paper. 7 There are many references in the literature to the so-called chicken-and-egg problem (who comes first the cardholders or the merchants?). See, for instance, Caillaud, B. and B. Jullien, 2003, Chicken & Eggs: Competition among Intermediaries Services, Rand Journal of Economics, vol. 34, No 2 (Summer), pp See Evans, D. and R. Schmalensee, 2005, The Industrial Organization of Markets with Two-Sided Platforms, NBER WP. Evans provides the following numerical example as an illustration of indirect network externalities. Consider a payment card system with 10 acquirers and 10 issuers, which would lead to 100 transactions. It is assumed that the market players have preferences such that the 10th (or marginal) acquirer/issuer has 1 Euro benefit per transaction, the 9th has 2 Euro, and so on, up to the 1st acquirer/issuer, which has 10 Euro benefit per transaction. It is straightforward to show that the per-transaction benefit received by the average acquirer is 5.5 Euro per transaction. The 10th acquirer gets only a 1*10 Euro from the overall transactions it engages in. But its decision to join the payment card system has benefited each of the 10 issuers. The 10th issuer only benefit as much as the 10th acquirer but everyone else benefits more. The 10 transactions generated by the last acquirer produce benefits on the issuer side averaging 5.5 Euro per transaction, for a total benefit of 55 Euro. That is, the indirect benefits to issuers are more than 5 times the direct benefits to the 10th acquirer. See Evans, D., 2002, The Antitrust Economics of Two-sided Markets, AEI-Brookings Joint Center for Regulatory Studies. 9 As the author himself notes, this correlation does not establish any causality between consumer usage and merchant acceptance. See M. Rysman, 2004, An Empirical Analysis of Payment Card Usage, Mimeo, Boston University. 6

27 In contrast with the indirect externalities, which become less and less important as the payment card systems mature, as already suggested by Katz (2001, op. cit.), Rochet considers that these fundamental externalities remain important even in mature payment card systems 10. Thirdly, payment card systems may, due to the existence of externalities between endusers, affect the volume of transactions by charging more to one side of the market and reducing the price paid by the other side by an equal amount: in other words, the price structure matters 11. The parameter via which platforms may affect the volume of transactions is the interchange fee. Raising the interchange fee would raise the merchant service charge but would lower cardholder fees. 2. What makes payment card systems different? Network externalities and complementarities between services offered to cardholders/issuers and merchants/acquirers are thus two important features of payment card systems. In this respect, POS payment card systems are significantly different from ATM (Automated Teller Machine) networks, because the latter is a one-sided market 12. The complementarities arise both on the cost side there may be economies of scale by having one firm produce multiple products and on the demand side there may be advantages in pricing complementary products together, mainly because this makes it possible to increase (and internalise) indirect network effects. In adopting a particular pricing strategy to cover both sides of the market, a payment card system would have to choose not only a price level but also a price structure for its service 13. In this regard, Rochet and Tirole (2003, op. cit.) consider that this theory is a cross between network economics, which emphasises such externalities, and the literature on (monopoly or competitive) multi-product pricing, which stresses cross-elasticities 14. There would be another factor that distinguishes payment cards from many industries, though certainly not all: the large fixed cost associated with establishing a viable payment network, which should lead to economies of scale above a certain output Economic modelling of payment card systems a survey The potential complexity regarding the role of various prices cardholder fees, merchant service charges and prices between customers and merchants has led to several recent formal analyses of payment card systems. Those analyses have mainly focused on the incentives for payment card systems when they choose interchange fees and have considered whether such fees promote a socially optimal choice of payment instruments. 10 Rochet, J.-C., 2003, The Theory of Interchange Fees: A synthesis of recent contributions, Review of Network Economics, Vol. 2, Issue See Rochet, J.-C. and J. Tirole, 2002, Cooperation among competitors: some economics of payment card associations, Rand Journal of Economics, vol. 33, No 4, Winter, pp , and Rochet, J.-C. and J. Tirole, 2003, Platform Competition in Two-Sided Markets, Journal of the European Economic Association, vol. 1, No 4, pp See Rochet, J-C. and J. Tirole, 2002, op. cit., for details. 13 See Rochet, J.-C. and J. Tirole, 2005, op. cit., for a formal discussion. 14 According to these authors, however, both from the positive and the normative viewpoint, two-sided markets differ from the text-book treatment of multi-product oligopoly, because the strong complementarities generate externalities that are not internalised by end-consumers, unlike the cases considered in the multi-product literature (the same consumer buys the razor and the razor blade). 15 Modern payment card systems may require large investments in communications and computing facilities in order to make card transactions convenient to cardholders and merchants and to minimise fraud. For a discussion of these issues, see, for instance, Evans, D., 2002, The Antitrust Economics of Two-Sided Markets, Joint Center AIB-Brookings Joint Center for Regulatory Studies. 7

28 Baxter 16 (1983) provided the first formal analysis of interchange fees in a payment card system. His analysis relies on three underlying assumptions: issuers and acquirers are perfectly competitive and make no profit, merchants do not accept cards for any strategic purpose (in particular, they do not accept cards to attract customers from rival merchants who do not accept cards), and, for working out the interchange fee under his analysis, it is implicitly assumed there is no variation in the benefits that merchants get from accepting cards. Schmalensee 17 (2002) relaxes the assumptions that issuers and acquirers are perfectly competitive and make no profit and that there is no variation in the benefits that merchants get from accepting cards. He emphasises the need to balance cardholder and merchant demand by setting an appropriate fee structure, although he does not derive cardholder and merchant demand from first principles. He shows that the privately and socially optimal interchange fees may coincide. Rochet and Tirole 18 (2002) also relax the assumptions that issuers and acquirers are perfectly competitive and make no profit and that merchants do not accept cards for any strategic purpose. They provide the first fully fledged model of an imperfectly competitive payment card industry, allowing a comparison between privately and socially optimal interchange fees. They explicitly model why competing merchants accept cards, and in so doing take account of the market interaction between consumers and merchants that arises. These authors are able to consider the full welfare effects of different interchange fees, allowing for the effects on cash-paying consumers as well. Finally, Wright 19 (2004) takes into account the heterogeneity of both consumers and merchants, and the imperfect competition between issuers and between acquirers. Unlike with Rochet and Tirole s model, in which merchants are homogenous and acquirers perfectly competitive, the socially optimal interchange fee involves a trade-off between giving consumers the right price signal for using cards and giving merchants the right price signal to accept cards. Privately optimal interchange fees may be too high, notably if merchant fees increase along with interchange fees but issuers do not pass the additional interchange fee revenue back to cardholders. In this case, high interchange fees are a way to transfer profits to that side of the scheme where they are least likely to be competed away, resulting in a restriction on output. 4. Effects of interchange fees Rochet and Tirole (2005, op. cit., p. 11) establish that the choice of an interchange fee has no real economic effect (i.e. is neutral) if the following conditions are jointly satisfied: first, issuers and acquirers pass the corresponding charge (or benefits) on to the cardholder and the merchant; second, the merchant can charge two different prices for goods or services depending on whether the consumer pays by cash or by card; in other words, the payment system does not impose a no-surcharge-rule as a condition for the merchant to be affiliated with the system; third, the merchant and the consumer incur no transaction cost associated with a system of double prices for each item. If one of these conditions is not observed, the interchange fee may no longer be neutral Baxter, W.F., 1983, Bank Interchange of Transactional Paper: Legal and Economic Perspectives, Journal of Law and Economics, 26, pp Schmalensee, R. (2002) Payment Systems and Interchange Fees, Journal of Industrial Economics, 50, pp See Rochet, J.-C. and J. Tirole, 2002, op. cit. 19 Wright, J. 2004, The determinants of optimal interchange fees in payment systems, Journal of Industrial Economics, Volume LII, pp For a more general model of the neutrality of interchange fees, see also Gans, J. and S. King, 2003, The Neutrality of Interchange Fees in Payment Systems, Topics in Economic Analysis and Policy, Vol.3, No 1. 8

29 The first condition seems to imply that the effect of a particular interchange fee depends on the market structure of acquiring and issuing and on the direct interactions between cardholders and merchants. For example, Carlton and Frankel (1995) demonstrate that if issuing, acquiring and merchant-level interaction are perfectly competitive then the interchange fee may have no impact on cardholder or merchant behaviour. This is because under perfect competition any change to the interchange fee is simply passed through all prices, including the prices paid by customers to merchants. These price changes offset each other so that altering the interchange fee has no real economic effect. Importantly, however, Wright (2004, op. cit.) shows that privately optimal interchange fees may be too high, particularly if merchant fees increase with interchange fees but issuers do not pass the additional interchange fee revenue back to cardholders. In this case, high interchange fees are a way to transfer profits to the side of the scheme where they are least likely to be competed away, resulting in a restriction on output. The second condition seems to imply that when a no-surcharge rule is effective, interchange fees can be used to have card users subsidised by those who do not use one (see Rochet and Tirole, 2002, op. cit., and Chakravorti and To 21, 2002). The reason would be that a rise in the interchange fee may lead to higher merchant service charges and, where surcharging is not possible, merchants may recover those charges from all customers and not simply those choosing to use a credit card. Rochet and Tirole, 2002, op. cit., in particular, demonstrate that even though this may increase merchant resistance to adopting card services a countervailing pressure arising from network effects may outweigh this resistance, leading to excessive card usage from a social perspective 22. The existence of transaction costs (third condition) seems to explain why even when surcharging is possible, thus allowing merchants to behave as a multi-product business and to charge prices contingent on the payment instrument used, very few merchants are tempted to do so (see Rochet and Tirole, 2003, op. cit.). According to the authors, transaction costs include the costs of drawing up a contact on how to pass through charges to the other side of the market and the costs of monitoring the actual transaction. 5. Competition between payment card systems The payment card industry is characterised by the existence of a few large players, which, according to Evans and Schmalensee (2005, op. cit.), is due to the existence of network effects and scale economies in the industry. A particular feature of this industry would be that competition between payment card systems may take place only on one side of the industry. Providing low fees or transfers to one side of the industry helps to obtain a critical mass of customers on this side, which in turn, due to network effects, encourages participation by the non-benefiting group on the other side 23. Competitive prices in a payment card system may however depend on the extent of multihoming on the other side of the market. Multihoming in the context of payment card systems describes a situation where most merchants accept cards from several payment card systems and many cardholders carry several cards. Rochet and Tirole (2004, op. cit.) consider that multihoming is an important countervailing force to the market power of card payment systems. For example, if payment 21 Chakravorti, S. and T. To (2003), A Theory of Credit Cards, Mimeo, Federal Reserve Bank of Chicago. 22 A merchant may have an intrinsic strategic interest in accepting payment cards in addition to the intrinsic benefits of receiving card payment rather than cash. Accepting card payment will increase the merchant s sales at the expense of other merchants. This strategic interest can be exploited by acquirers with market power, or by issuers through an interchange fee, as shown by Rochet and Tirole. The additional surplus that acquirers are able to extract will therefore lead to higher consumer prices. 23 See Caillaud, B. and B. Jullien, 2003, op. cit. 9

30 card system A reduces its interchange fee and this decrease is partially passed on in the MSC paid by the merchant, the latter may be more tempted to turn down a costlier competitor of A where a large proportion of this competitor s cardholders also carry cards issued by A. For this reason, Rochet and Tirole (2004, op. cit.) claim that multihoming intensifies price competition between payment card systems. However, multihoming will not play this countervailing role on the market if acquirers charge the same fees to merchants for accepting credit and debit cards issued by different networks ( blending ). Moreover, even where blending is not practised, it should also be noted that multihoming will be widespread only if the fixed costs of joining a payment card system are absent or low. 6. Competition among means of payment Modern economies have several types of payment instrument at their disposal, e.g. cash, debit and credit cards and e-purse cards. Since banks are likely to offer simultaneously all types of payment instrument to their clients (cardholders and merchants), it is therefore relevant to analyse the role of the prices and associated costs of these payment instruments from the viewpoint of the bank s overall profitability and from a welfare perspective. In particular, it is of interest in the context of the sector inquiry to assess whether (1) credit or debit cards are relatively less costly than other means of payment and (2) the prices of these means of payment do in fact reflect their associated costs. If credit and debit cards are less costly than other means of payment offered by the same institution, but their prices do not reflect this difference in costs, this might imply that credit and debit cards are cross-subsidising other means of payment. Moving towards a more efficient means of payment would therefore reduce the importance of the role played by the interchange fee as a cross-subsidising factor among different payment instruments. If this is the case, one may also ask whether the merchant side is currently not relatively overpriced. There is some empirical literature on competition among means of payment. Guibourg and Segendorf (2004) 24 estimate the private costs incurred by banks to provide different payment services and investigate to what extent the price structure reflects the cost structure. This study uses data on the prices, which are mainly of a two-part tariff nature and include MSCs, and costs (fixed and variable) of each payment instrument offered by the four largest Swedish banks, which together account for 92% of the card and credit transfer market and 96% of the cash distribution market. The data are from The main hypotheses to be tested are whether (i) transaction fees equal average variable costs (used as a proxy for marginal costs), (ii) relative prices reflect relative costs for some payment services considered as substitutes, and (iii) there are cross-subsidies among the payment instruments. Several important findings were made by the authors. Firstly, there are considerable differences in costs between payment instruments. Paper-based payments are more costly to process than electronic payments and debit card payments are less costly than credit card payments and cash withdrawals. Secondly, the hypothesis that transaction fees equal variable costs is refuted for several types of payment instruments, notably as regards the acquiring of credit and charge cards, credit transfers and ATM withdrawals. Thirdly, almost no information about these relative costs is passed on to consumers through price mechanisms. In the POS market, relative prices give no information to support a cost-efficient choice between debit cards, credit cards and ATM withdrawals. Here, the debit card has a large cost advantage not 24 Guibourg and Segendorf, 2002, Do prices reflect cost? A study of the price and cost structure of retail payment services in the Swedish banking sector. 10

31 reflected in private variable fees 25. Fourthly, and most important, all means of payment except for cash distribution gives rise to net revenues for the banks, where the most net revenue is generated by card payments (acquiring and issuing). Therefore, the non-transparent price structure of payment instruments (with few exceptions consumers do not pay any variable fee and receive insufficient price signals to make their choice) requires cross-subsidising between them. If banks were to move towards a more cost-based price setting strategy, this would result in the introduction of transaction fees for ATM withdrawals, with a consequent decrease in the fees for acquiring services. The reason why banks do not implement more cost-based pricing strategies is because they are locked into a prisoner s-dilemma type of situation. While every bank would gain from a joint shift to costbased pricing, it may be very costly for a bank that goes it alone 26. From a welfare perspective, and given that user demand is price-elastic, a pricing strategy based on variable costs would lead to greater use of debit cards, less use of credit cards and cash, and more use of electronic credit transfers and direct debits. Such a change in the choice of payment instruments would lower costs for the banks involved. Banks could reduce total production costs and probably increase overall profitability by adopting more cost-based pricing strategies that would re-orient demand towards less costly payment services. A follow-up of this study was conducted by Hans Brits and Carlo Winder (2005) in order to quantify the social costs of certain payment instruments (cash, debit card, e-purse and credit card) in the Netherlands 27. The cost data used in the study came from four major banks, with a joint market share of about 90% in the Netherlands, and the Netherlands s central clearing organisation, Interpay, and were for As well as the hypotheses tested in the study of Guibourg and Segendorf (2004), it also considers the break-even transaction amounts, i.e. those amounts for which the costs of two payment instruments are equal, and provides a breakdown of costs among the market participants. Several important conclusions may be drawn from this study. Firstly, and confirming the results obtained for Sweden, the lowest average costs per transaction were, in ascending order, on e-purse payments, debit card payments, credit card payments and cash payments. The credit card is less attractive than the debit card, irrespective of the transaction amount, as the credit card s variable costs both for the execution of the transfer itself and in relation to the size of the purchase are larger. The e-purse is cheaper for any transaction amount. Cash is more economical for purchases below a certain threshold. Secondly, the costs related to cash payment are largely incurred in the retail sector, whereas a relatively large share of the costs relating to electronic POS payments is borne by the banking industry. Following Guibourg and Segendorf, the authors conclude that the current tariff structure of payment instruments in the Netherlands does not provide sufficient incentives to stimulate the use of the most efficient payment instrument 28. Due to the non-transparency of tariff structures, large-scale cross-subsidisation is likely to be present in the Netherlands as well. The authors also claim that the reason why banks do not implement a more cost-based pricing strategy is because they are locked into a prisoner s-dilemma type of situation. 25 Interestingly, the results also show that the pricing for private costumers appears to reflect costs less than the pricing of payment services for corporate costumers. 26 Consumers may also incur a transaction cost when switching from one type of means of payment to another, which may difficult the implementation of such a strategy by banks. 27 Brits H. and C. Winder, 2005, Payments are no free lunch, De Nederlandsche Bank, Occasional Studies, vol Interestingly, results show that pricing for private costumers appears to reflect costs less than the pricing of payment services for corporate costumers in the Netherlands as well. Cardholders are confronted with costs in the form of annual fees for e.g. debit cards, while merchants factor their costs into the prices of services. 11

32 It is interesting to note that similar results were found in a recent replication of this study in Belgium (Guy Quaden, 2005) 29. Similar policy implications as regards the role of correct price signals for the choice of the most efficient means of payment can thus also be drawn for Belgium. Finally, Bolt, Humphrey and Uittenbogaard (2005) 30 use the experience of Norway and the Netherlands over to try to determine what the incremental effects of transaction pricing may be on the use of debit cards versus withdrawing cash from an ATM and on the use of electronic giro transactions (credit transfers and direct debits) in preference to paper transfers. Specifically, the study compares the use of payment instruments per person in Norway, in response to the prices charged and the availability of terminals, and the level of actual consumption with the example of the Netherlands, which also adopted electronic payment but did not price. The cases of Norway and the Netherlands are interesting because the first country priced its payment services and the second one did not. In fact, Norwegian market players coordinated the timing of when the direct pricing of consumer payments would start, but not the level of prices to be charged. The result was to eliminate the bank practice of recouping payment costs through payment float debiting consumer accounts prior to the value date for bill payment or delaying crediting to accounts. They contrast the rapid adoption of electronic payments in Norway over with the experience of the Netherlands, which also rapidly adopted electronic payment but did not impose per-transaction prices on consumers. If the incremental effect of direct pricing is large, holding constant within the cross-country influences affecting the adoption of electronic payments, then the potential social benefit can also be large. Results show that the effect of terminal availability on relative debit card and ATM use exceeds that of pricing. Even so, pricing has a significant effect in influencing payment choice, though not as much as expected, because consumers mainly look at the convenience, safety and other non-price attributes of different payment instruments. The overall conclusion is that while terminal availability appears to have a stronger effect on the relative use of payment instruments than direct per-transaction pricing, the shift to electronic payments could be boosted where pricing is combined with terminal availability. Given that electronic payments are considerably cheaper than their paper-based alternatives (including cash), banks and merchants have an interest in shifting users to electronic payments to save costs. Naturally, this would reduce the importance of the role played by the interchange fee as a cross-subsidising factor among different payment instruments. It also seems from this that currently the merchant side is relatively overpriced where debit card payments are concerned, and that the use of electronic payments can be further stimulated by reducing merchant service charges (and directly charging cardholders). 29 Quaden, G., 2005, Coûts, avantages et inconvénients des différents moyens de payement, Banque Nationale de Belgique. In Belgium, debit cards are less costly than credit cards, credit cards are less costly than cash and cash is less costly than Proton (e-purse). More precisely, cash and debit cards are have the same costs for transactions below euros but above this amount the debit card is the cheaper means of payment. 30 Bolt, Humphrey and Uittenbogaard, 2005, The Effect of Transaction Pricing on the Adoption of Electronic Payments: a cross-country comparison. 12

33 III. Data and Methodology 1. Data issues DG Competition has collected information from two main sources. Information on acquiring and issuing was collected through a questionnaire sent out in July 2005 to a representative sample of 203 acquirers and issuers. Data on payment card networks was obtained from a questionnaire sent out in August 2005 to 26 domestic and international payment card systems. Following this initial request, DG Competition sent out an additional questionnaire to payment card networks in December The geographical scope of both questionnaires was the EU-25. In addition, DG Competition also had at its disposal surveys of consumers and merchants behaviour and a range of further market studies 31. As regards the type of products and services covered by the questionnaires, it should be noted that the questions put to acquirers and issuers addressed only debit and credit cards (deferred debit cards were treated as credit cards) 32. Moreover, they focused only on transactions made at physical points of sale (POS) and did not cover ATM transactions. In contrast, the questionnaire sent to payment card systems covered a wide range of rules and activities developed by these institutions as well as the relevant price and cost data. Prior to the questionnaires, DG Competition held consultations with stakeholders such as industry associations and consumer groups. It is also important to point out that the questionnaires were previously road-tested with a small group of market players to make them operational and iron out errors, so that respondents would not expend unnecessary resources in gathering and supplying the information. To further facilitate the process, all questionnaires were distributed in electronic format. Finally, DG Competition communicated with the addressees and provided detailed answers to any outstanding questions. The information was collected at institution and country level. If the selected issuer or acquirer was a parent company, data were gathered separately for each member (subsidiary) entity if these entities followed different pricing policies 33. Moreover, where institutions were active in more than one EU Member State, the information was obtained separately for each country of operation. Finally, information was collected separately for each payment card system. Similarly, payment card systems provided information separately for each country if they had activities in different EU countries. The information was mainly collected on a yearly basis and over the period Some data, however, were collected on a quarterly basis. A significant amount of the requested information concerned financial aspects (e.g. prices and costs). In order to harmonise the financial data, respondents were asked to convert their data into euro currency. To this end, DG Competition provided a table with the average annual ECB exchange rates for the reference year to convert the requested annual data and with the average quarterly ECB exchange rates for the reference quarters to convert the requested quarterly data. For the same questions, different respondents provided data expressed both in euros and as a percentage of another variable. For instance, some institutions reported prices in euros per transaction while others reported them as a percentage of the transaction volume. In order to 31 It includes, for example, the ABR report on payment cards. DG Competition also consulted seminal economic literature on the payment card industry. 32 Store cards (cards issued by non-banking institutions for use for payment in specified stores) were explicitly excluded from the scope of the questionnaire. 33 If the questionnaire was sent to a subsidiary entity, the entity was instructed to forward it to the parent company. 13

34 have a meaningful comparison across institutions and Member States, these figures were converted into euros or percentages as appropriate. Some of the requested data required an allocation of revenues and costs based on accounting data. This allocation was made by the respondents themselves. Finally, some data were collected in nominal terms while others were collected as weighted averages. When weighted averages were requested, the questionnaire provided the relevant weights. The response rate was close to 95% for the questionnaire addressed to acquirers and issuers and virtually 100% for that addressed to payment card systems. 2. Methodology The data set covering acquirers and issuers has been collected and assembled by DG Competition and DG JRC, ISPRA (VA), Italy. Once assembled, the final database was submitted to statistical tests in order to identify possible outliers (i.e. extreme observations) that could have created a significant bias in the analysis. Figures diverging significantly from the mean of the overall and country samples were thus identified. In many cases, DG Competition had to contact the respondents again to check the correctness of the reported figures. A significant number of initially detected outliers were thus corrected. However, in cases where no evident explanation for an outlier was found, it was simply dropped from the data set. The final data set is unbalanced, because some respondents reported figures only for a few years and not over the entire time period. There are significantly more observations reported for later years (2004). Where relevant, both for economic and statistical reasons, only a balanced data set was considered in the analysis. Moreover, some respondents did not provide data on all questions of the questionnaire. The cited reasons related to technical limitations and a data reporting model making it impossible to provide data with the required detail. Consequently, a high response rate for the overall questionnaire did not necessarily imply that all questions were addressed in an equal and full manner. The analysis carried out in this inquiry used both descriptive statistics and econometric techniques. Descriptive statistics included simple and weighted averages, minimum and maximum values, and standard deviations. These statistics were computed mainly at country level using both a cross-sectional (i.e. at a single point in time) and a time-series analysis. The cross-sectional analysis usually used the last year of the period (2004). The econometric techniques included the standard ordinary least-squares estimator and some sophisticated paneldata techniques such as the between-effects estimator, the fixed-effects estimator and the random-effects estimator. The econometric exercise aimed to identify the main determinants of some of the prices in this type of industry and it was performed by DG JRC, ISPRA (VA), Italy. 3. Sampling 3.1 Technique The selection of addressees was possible only for credit and debit cards. For debit and credit cards, separate selection was done for issuing and, to some extent, acquiring to avoid a possible bias. The selection of issuing institutions was primarily based on total cardholder expenditure volume for each respective institution. Where, for various reasons, cardholder expenditure volumes were not available, a proxy in the form of the number of cards issued by each respective institution was used. Later, the report provides a justification for using this kind of proxy. After separate selection of issuing and acquiring institutions for both debit and credit cards, the lists of institutions obtained were merged. A detailed verification of the final list was 14

35 then performed. In some exceptional cases, when a previously unknown fact came to light (e.g. merger of two selected undertakings), the institution was removed from the list. The selection or sampling was performed on a list of banks submitted by the two biggest (at EU level) international payment networks: MasterCard and Visa. The financial data for both networks was merged to create a single list. In the cases of Austria and Belgium, the necessary data on cardholder expenditure volume were available only for groups of members jointly holding licences for the payment card networks. Here, a proxy in the form of the number of cards issued was used instead; however, these figures were available only for MasterCard members. The correlation analysis performed revealed that, although positive, the correlation between the number of cards issued for MasterCard and cardholder expenditure volume on these cards was nonetheless not overly significant, amounting only to 67%. However, the subsequent correlation analysis comparing the number of cards issued for the MasterCard network and total cardholder expenditure volume, i.e. the volume registered on both respective networks, yielded an astonishing 89%, leaving practically no doubt that the selected proxy worked. Two different techniques were used to select institutions active in issuing and acquiring. These techniques were the same for both debit and credit cards. The main selection was from issuing undertakings due to the nature of the issuing business. In fact, the evidence shows that there are many more issuing than acquiring institutions across the European Union. This is partly due to the rule in the past, enforced by certain payment card networks, that an institution had to be an issuer before it could apply for an acquiring licence, and partly due to the fact that issuing is considered to be an intrinsically more profitable business than acquiring. The data set comprised a list of issuing institutions sorted by countries. The sampling was done on a per-country basis, as a random selection could have led to the underrepresentation of some EU countries in the sample (particularly small ones). First, the total number of issuers per country was computed to assess whether sampling for a given country was needed. In countries with fewer than 8 issuing institutions, no additional selection was done. Their selection was automatic. For other countries, the principle of proportionality was applied to ensure that the country size was represented in the final sample, i.e. the number of institutions to be sampled was reduced in proportion. Second, issuers were sorted within the country in descending order according to their share of the total cardholder expenditure volume of all issuers in the country. If random sampling had been applied to the whole dataset, the largest player(s) with significant market share might not have been included in the sample. In an attempt to avoid this while nonetheless preserving an element of randomness, the following technique was used. The issuing undertakings with a share of 50% were included in the sample automatically. However, selection was random for the rest of issuers in the data set. The remaining issuers were split into 3 groups according to the 33rd and 67th percentiles, using the cardholder expenditure volume. From each group, an equal number of institutions were randomly selected, with the combined total equal to the proportionally reduced number to be sampled (see previous paragraph). Third, the list of selected banks in the sample was extended to include the largest acquiring institutions which for some reason were not active in issuing (e.g. in Germany, Austria, etc.). Nonetheless, given that the majority of acquirers in general are also issuers, additional detailed sampling for acquiring was not needed. If this had been done, the result could have been a sample as big as the original pool. Next, the procedure was repeated for institutions active in debit cards. Prior to selection, the list of issuers of debit cards was compared to that of issuers of credit cards, and it was found 15

36 that the lists differed significantly, particularly in certain countries where the biggest issuer of debit cards was not engaged in credit card business. Therefore, additional sampling was needed. Finally, the samples for debit and credit cards were merged to create the final sample of addressees. This list was thoroughly verified and finalised. The final list contained 232 institutions. 3.2 Main drawbacks of the technique The technique, as it stands, has a number of statistical defects, which were carefully assessed prior to sampling. These are as follows: 1) Large institutions are intrinsically over-represented (does not meet the criterion of equal representativeness of different-size banks in the sample). Explanation and justification: There is a double bias towards large institutions. Firstly the bias was created when the 50% criterion was introduced, and secondly it was reinforced when the list was extended to include the largest acquiring institutions. This bias represented a conscious sacrifice of degrees of freedom for the sake of including the most important players. Despite the fact that the questionnaire was intended to address a representative sample, the influence of the biggest European players cannot be ignored. Moreover, there may be a higher risk with competition problems arising among the biggest players (with important market shares) rather than among small ones. 2) There is a strong bias towards MasterCard and Visa network members. Explanation and justification: This bias was allowed given that the majority of MasterCard and Visa networks are also members of other networks, namely national domestic schemes, American Express and Diners Club. This bias can thus be considered to be of no particular pertinence and will therefore not affect statistical representativeness in any significant way. 4. Data patterns For merchant service charges, the final data set contained information from 147 acquiring institutions operating in the 25 EU countries. This resulted in a total of unique bank/network/year combinations. The number of combinations increases significantly if the data are disaggregated at merchant level ( unique merchant/bank/network/year combinations). The distribution of the total combinations over the period varies significantly. On average, the number of observations in 2004 is more than double of that in The distribution of the observations across the EU-25 Member States also varies considerably. For example, in 2004 (the period with the highest number of observations) Poland has 1 bank/network/year and 60 merchant/bank/network/year combinations, respectively, while Spain has 28 and Poland and Spain are thus, respectively, the worst and the best represented countries in the sample. Finally, the two most important international payment networks (MasterCard and Visa) accounted for significantly more observations compared to both other international payment networks (American Express and Diners Club) and national payment networks. 16

37 Section B Financial Aspects of the Industry

38 IV. Interchange Fees In this chapter, we analyse an important driver of pricing in the card payments industry: so-called interchange fees. These fees are paid between banks and determine to a large extent the final prices paid by merchants for payment card acceptance. The setting of interchange fees and fee levels are important issues for competition in the European payment cards industry, because it appears that: there is significant dispersion in the level of interchange fees across EU Member States; interchange fees can help to determine the level of other important fees in the market: notably acquiring banks fees to businesses; interchange fees can permit a significant transfer of revenue from acquiring banks to card issuing banks; and major players in the payment cards industry and recent academic research argue that the interchange fee is an efficient and necessary instrument for allocating costs and revenues in a POS card system. We subsequently: examine the rationale for using interchange fees in a card payment system; analyse how interchange fees are set in the international and domestic systems; compare interchange fees in the two large international systems (MasterCard and Visa) with regard to fee structures, nominal levels and weighted average levels; and compare interchange fees in national systems with regard to nominal levels and fee structures. 1. Purpose of interchange fees 1.1 Academic explanations As explained in chapter 2, most economic analyses of payment card systems have focused on the incentives for payment card systems when they choose interchange fees. Several studies have considered whether such fees promote a socially optimal choice of payment instruments. Rochet and Tirole 34 (2002) compare privately and socially optimal interchange fee levels. They consider the full welfare effects of different interchange fees, also allowing for the effects on cash-paying consumers. Wright (2004) finds that privately optimal interchange fees may be too high, notably if merchant fees increase with interchange fees but issuers do not pass the additional interchange fee revenue back to cardholders. Where this rebate is not provided, high interchange fees may have the effect of transferring profits to the side of the scheme where they are least likely to be competed away, resulting in a restriction on output. To summarise therefore, two competing assessments can be distilled from the economic literature on interchange fees in payment card systems: either that their effect is neutral and provides efficient incentives for card issuers to expand output; or that high interchange fees offer a means of transferring rent (which cannot be competed away) from acquiring to issuing banks. 34 See Rochet, J.-C. and J. Tirole, 2002, op. cit. 17

39 1.2 Industry explanations The Commission asked international and domestic payment card networks to explain the economic function that interchange fees fulfil in their networks. The networks were also asked to provide the market context (for instance, in terms of the mix of different payment means, the maturity of the payment card segment, or regulation) explaining why interchange fees were used or not. 35 One of the international payment card systems believes that in the absence of POS interchange fees paid by acquirers to issuers, issuers would have to recoup all of their costs from cardholders and this would lead to a level of card issuing not optimal for the system as a whole. The other system identifies an imbalance between issuing and acquiring costs necessitating a transfer of revenues from acquirers to issuers. The common feature of the MasterCard and Visa replies seems to be that both networks assume that issuing banks would not gain sufficient revenue from issuing MasterCard/Visa cards in the absence of POS interchange fees. (This assumption is challenged empirically in chapter 7 of this report.) On the basis of this assumption, MasterCard and Visa assert that total system output would fall if card issuers were not subsidised through a transfer of revenues from acquirers. Domestic payment card networks often did not explain why POS interchange fees are used in their system, but referred to a declaration of banks represented on the European Payments Council. This declaration stated that interchange fees have proven to be necessary enablers for the operation and development of the cards business and for sound cooperation between competing banks. Amongst those domestic networks that commented on the purpose of POS interchange fees, opinions diverge as to the character of such interchange fees. Some view interchange fees as remuneration for services provided by issuing banks to acquiring banks, which appears similar to the purpose of interchange fees in an ATM system. Such networks accordingly often set interchange fees on the basis of the costs of the services that the participating banks provide to each other. Other networks, however, reject the idea that POS interchange fees could be a price and argued that POS interchange fees are a tool to shift costs and revenues in a way that is neutral in terms of the overall costs and revenues incurred/charged by the banks in the system. One scheme owner argues that its interchange fees are set to foster the contribution of the [..] system in reducing payments via means typically costing more or that are less efficient for the banking system, such as cash or cheques. This implies that the level of interchange fees might be set to encourage greater use of efficient payment means. To summarise, a substantial number of networks maintain the traditional view that POS interchange fees remunerate services that banks provide to each other within the network and compensate for corresponding costs. This is reflected in the fee-setting practice of a number of networks. Other networks appear to have adopted views advanced more recently by academic authors. 2. Setting of interchange fees 2.1 International schemes In the international MasterCard and Visa systems, member banks or representatives of member banks usually set the POS interchange fees. In at least one of these systems, however, a mixed methodology for setting domestic interchange fees was observed: in country A, domestic interchange fees are set by a local board of member banks ( multilateral interchange fees ), in networks replied to the question as to the purpose of POS interchange fees and three indicated a specific purpose for ATM interchange fees. Where networks specifically commented on the purpose of ATM interchange fees, the reasons diverged from the reasons given for POS interchange fees. 18

40 country B by the management of the scheme owner, and in country C between pairs of banks in bilateral agreements ( bilateral interchange fee agreements ). In the other international system, the second option is not used. As already noted in the Commission Decision on Visa cross-border interchange fees 36, at least one of the international systems has created an order of precedence for cross-border and domestic interchange fees (a fallback interchange fee system). First, bilaterally agreed fees always take precedence over multilateral interchange fees set by a board of member banks. This holds both for cross-border and for domestic fees. Second, where banks neither bilaterally nor multilaterally agree on the level of domestic interchange fees, multilaterally set cross-border interchange fees will apply by default to domestic payment card transactions as well. Three points merit attention in this context: The effect of the fallback interchange fee system appears to be that in the absence of an agreement between member banks, there will always be an interchange fee that acquirers pay to issuers, whether a multilaterally agreed default rate at local level or a multilaterally agreed cross-border fee; this excludes the possibility that acquirers pay no interchange fees to issuers. At domestic level, local member banks of an international system may charge each other preferentially low fees (on the basis of bilateral agreements) while applying higher fallback rates (set by a local board of banks) to foreign banks or other outsiders that attempt to compete with them. Where local boards of banks multilaterally set domestic interchange fees in the international systems, they appear to view cross-border interchange fees as a minimum benchmark for setting these domestic fees. 2.2 Domestic schemes From the replies received, there appear to be four different means of setting interchange fees (where interchange fees are set at all). First, the scheme s management sets interchange fees without the intervention of member banks. Second, member banks bilaterally agree on interchange fees. Third, member banks multilaterally agree on interchange fees. Fourth, member banks multilaterally agree on a fee paid by merchants to processors, who collect this fee and then transfer it to the appropriate issuing bank without the involvement of an acquiring bank. The last system is unique to one Member State (Germany). 3. Level of interchange fees 3.1 Level of interchange fees: cross-border interchange fees This section describes the interchange fee structure and compares cross-border interchange fees on a nominal and weighted average basis. It will be shown that weighted averages provide a more meaningful comparison of the general level of interchange fees applied in a network Fee structure a) Distinction between MC/Visa cross-border and domestic fees As already noted in the Commission Decision on Visa cross-border interchange fees 37, at least one of the international systems distinguishes between domestic, intra-regional and inter- 36 Commission Decision of 27 July 2002, OJ L 318/17 of 22 November 2002, pt Commission Decision of 27 July 2002, OJ L 318/17 of 22 November 2002, pt

41 regional interchange fees. Domestic interchange fees are defined as transactions in the same country in which the card is issued. Intra-regional interchange fees (hereafter referred to as cross-border interchange fees ) apply to transactions that are completed at a merchant outlet outside the country (but within the geographical region) in which the card is issued. Interregional interchange fees apply to transactions between Europe and Asia or the US; these fees will be disregarded in what follows. b) Distinctions within MC/Visa cross-border interchange fees Cross-border interchange fees may differ according to the method of processing (on-line, off-line, card present/not present etc.) and the type of card used (consumer or commercial cards; magnetic stripe card or chip card, etc.). Specific cross-border interchange fees for individual merchant segments are the exception: only one of the international systems has a specific interchange fee rate for cross-border transactions in the airline sector Comparison of nominal fees (MC/Visa, 2005) a) Credit cards As is apparent from the Visa website 38, Visa Europe s cross-border (intra-regional) interchange fee scale has many different categories for consumer credit cards alone. Interchange fees range from 0.65% for consumer transactions processed with the EMV chip up to 1.05% for non-electronic transactions. MasterCard s intra-european interchange fee 39 scale has ten different categories for consumer credit cards alone. Interchange fees range from 0.8% for consumer chip transactions up to the higher base rate of 1.3% (standard consumer) and 1.9% (World Signia) on consumer transactions. Unlike Visa, MasterCard also publishes its crossborder interchange fees for commercial card rates, which range from 1.2% to 1.90%. b) Debit cards Interchange fees for transactions with credit cards are denoted for both networks in terms of a percentage of the transaction value, while interchange fees for transactions with debit cards are denoted differently for each network. Visa applies a fixed fee per transaction, whereas MasterCard opts for a percentage of the transaction value. Visa Europe s Intra-regional Interchange Fees for Debit Cards range from 0.27 to 0.30 for non-electronic transactions. MasterCard s Intra-European Interchange Fees for Debit Cards operate on a percentage basis, ranging from 0.50% for chip transactions up to 1.15% for e-commerce transactions. A comparative analysis of the effect that the two different charging mechanisms have on a network s revenue stream is given below in the next section. The analysis shows how, for average transaction values, both pricing formulas produce similar interchange fee levels Comparison of weighted average fees (MC/Visa) A comparison of interchange fees based on nominal rates alone does not yield results that fully represent the levels of interchange fees applied in the market. Focusing solely on nominal fee (or fee-tier) levels would not take into account the frequency with which different tiers are applied in practice, i.e. the respective merchant sales volumes for each tier. Therefore, a weighting exercise was carried out. Weights were calculated according to the respective turnovers for each respective interchange fee tier per EU-25 Member State (see methodological details in Annex 1)

42 a) Credit cards Average cross-border fees across the EU for credit cards are shown below in Graph 4. Graph 4 EU Weighted Average Cross-Border Interchange Fee - Visa vs MC, , % 1,30% 1,25% MC Visa 1,30% 1,25% 1,20% 1,20% IF, % 1,15% 1,15% 1,10% 1,10% 1,05% 1,05% 1,00% Year 1,00% The comparison highlights two important findings: Before 2001, Visa acquirers appeared to be paying on average somewhat higher interchange fees for cross-border transactions than MasterCard acquirers. As of 2001, MasterCard interchange fees started to exceed Visa interchange fees and continued to do so up to 2004, the end point of the analysis. In general terms, over MasterCard acquirers seemed to pay on average 6% higher interchange fees on credit card transactions across the EU- 25. Credit card fee rates for the two networks seemed to be more or less similar until 2001, but from 2002 the two rates started diverging strongly. In , the absolute difference in interchange fees paid by MasterCard and Visa acquirers amounted to about 1-2%. By 2002, however, the average difference between the two fee levels already amounted to about 10%. Throughout the whole period from 2000 to 2004, MasterCard cross-border interchange fees kept rising (up to 2004, when they fell only insignificantly), while Visa rates followed a steady falling trend. One of the most likely explanations for falling Visa rates is the adoption by the European Commission in 2002 of the Visa Decision 40. This decision fixed the underlying cost components for consumer card interchange fees and obliged Visa to conduct an in-depth cost study to justify the level of each of the costs. Moreover, the Decision set an annual ceiling on the interchange fee rates for each subsequent year up to This appears to have had the effect of reducing Visa cross-border interchange rates. MasterCard cross-border rates remained unregulated, which allowed the network to keep interchange fees significantly above the rates of Visa. 40 OJ Press Release of 24/07/2002, reference IP/02/

43 b) Debit cards Average cross-border fees across the EU for debit cards are shown below in Graph Graph 5 IF, % EU Weighted Average Cross-Border Interchange Fee - Visa Debit/Electron vs Maestro, , % 1,000% MC 0,900% Visa 0,800% 0,700% 0,600% 0,500% 0,400% 0,300% 0,200% 0,100% 0,000% Year 1,000% 0,900% 0,800% 0,700% 0,600% 0,500% 0,400% 0,300% 0,200% 0,100% 0,000% The patterns observed in cross-border interchange fees for debit cards mimic to some extent the patterns observed in cross-border fees for credit cards. Over , the average weighted interchange fees paid for debit card transactions proved to be somewhat higher for Visa debit, whereas from 2002 Maestro cross-border rates started exceeding those of Visa debit. The comparison shows that: Visa debit card interchange fee rates were rising up to However, the average fell significantly by about 25% following the 2002 Visa Decision. In 2004, however, Visa EU-25 weighted average interchange fees rose again somewhat, possibly due, among other things, to more extensive use of more expensive interchange fee tiers, i.e. higher turnovers in these tiers, with consequently a higher average fee level. Maestro rates continued rising up to 2004, when there was a slight drop. Over the whole period, MasterCard acquirers paid on average 12% higher interchange fees on debit cards than Visa acquirers. Furthermore, due to the sharp fall in Visa cross-border debit interchange fees in 2002, the spread between the two networks fees widened in In 2004, however, due to slight drop in Maestro fees and a rise in Visa debit card fees, the gap somewhat narrowed. Analysis of the fee structure for transactions with MC/Visa debit cards: Visa acquirers pay a fixed per-transaction fee, while MasterCard levies a percentage of the transaction value. It may happen that one fee mechanism can generate higher returns than another. A fee structure analysis shows that for an average transaction value (ATV) below 49 euros, the Visa fee mechanism generates higher interchange revenues than the MasterCard fee mechanism, while the converse is true for an ATV above 49 euros (see Annex 2). For actual transaction values, 41 The level of the weighted average cross-border interchange fee charged for Maestro and Visa Electron transactions may be subject to an upward bias created by the level of interchange fee reported by a Danish acquirer. The Danish interchange fee was substantially higher than that in other countries. However, since this applies in an equal manner to both the MasterCard (Maestro) and Visa (Visa debit) networks, the relationship and perceived differences in the level of the weighted average cross-border interchange fees between the two networks still holds true. Furthermore, this does not affect the overall trend in fees across networks over the time period examined. 22

44 however, it appears that both fee mechanisms generate similar fee revenues, with Maestro slightly above Visa debit. This is shown below in Table 1. Table Maestro average transaction value Maestro rate (%) 0,55% 0,55% 0,55% 0,55% 0,55% Maestro fee per transaction 0,270 0,308 0,281 0,275 0,275 Visa debit fee per transaction 0,27 0,27 0,27 0,27 0,27 Difference in fee - 0,001 0,038 0,011 0,005 0, Level of interchange fees: domestic interchange fees for credit cards Fee structure MasterCard and Visa domestic interchange fees differ according to the method of processing and card types. Unlike cross-border interchange fees, specific rates for merchant segments are common for domestic interchange fees across the EU. This distinction according to merchant segment is most evident in two Member States, where member banks in both networks even have different interchange fees for individual merchants. Merchant-specific interchange fees are more common for Visa cards (in 24 of the EU-25 countries) than for MasterCard cards (in 10 of the EU-25 countries). Petrol stations and airlines regularly have specific rates Comparison of nominal fees for credit card transactions (MC/Visa) The analysis of domestic interchange fees for MasterCard and Visa credit and charge cards shows that the fee levels tend to remain static over time. Country divergences as regards domestic interchange fees (nominal rates) in the MasterCard and Visa systems are considerable, both for credit and debit cards. Taking the minimum rates as a benchmark, the nominal rates of Visa debit card interchange fees in 2004 diverged across the EU by as much as 220%. Similarly, domestic interchange fees for Maestro cards diverged by 280% across the EU. The picture is similar for credit and charge cards. The level of domestic interchange fees for Visa cards diverged by as much as 323% across the EU and by 329% for MasterCard cards. In most Central European countries, the nominal rates for MasterCard and Visa credit and debit cards are set at identical levels. In many of these countries, banks do not appear to have specific product features for individual card brands or distinguish between credit and debit fees. As noted previously, domestic interchange fees in the MasterCard and Visa systems regularly distinguish between merchant segments. Three important conclusions can be drawn from the comparison of MasterCard and Visa domestic interchange fee rates. First, there are considerable differences between interchange fees from one merchant segment to another. Acquirers in the same country may pay roughly half the interchange fee for credit card payments at a petrol station than for a credit card payment to an airline. Second, country-specific differences in a given merchant segment are considerable as well. For instance, Portuguese acquirers bear roughly 165% more interchange fee costs for a credit card transaction at a petrol station than their German counterparts. Third, merchant-specific fees within one and the same country and the same merchant segment also differ to some extent between MasterCard and Visa. 23

45 Fourth, according to an industry expert, some large merchants in specific sectors (for example, the petrol and food sectors) may benefit from special interchange fee rebates. Thus, while the same interchange fee rate applies to all transactions of merchants in the same sector, some merchants may have part of the interchange fee retained by the acquirer to be later transferred back to the merchant, thus de facto reducing the actual merchant fee Comparison of weighted average fees for credit card transactions (MC/Visa) a) EU-wide comparisons Interchange fees on credit card transactions have been compared for MasterCard and Visa, weighted by tiers of domestic turnovers and averaged out across the EU 25. The comparison has led to the following observations (see Graph 6 42 ): Since 2000, Visa domestic weighted average interchange fees have fallen gradually, while MasterCard interchange fees show no distinct trend, making it difficult to draw any precise conclusions. By 2004 (over the 5-year horizon), Visa domestic interchange fees had fallen by around 5.7%, while MasterCard fees fell by 6.5% The weighted average domestic fee level was on average somewhat higher in the MasterCard network. Over , MasterCard charged on average 3% higher domestic interchange fees on credit cards than the Visa network. Domestic interchange fees appeared to be converging up to In, 2002, they diverged significantly (by 5%), but from then on they started converging again, reaching very close values by 2004 with a reported difference of about 1%. While Visa domestic interchange fees were dropping throughout the whole period, the trend in MasterCard weighted average domestic interchange fees was unclear. Graph 6 EU Weighted Average Domestic Interchange Fee - Visa vs MC, , % 1,25% 1,20% MC Visa 1,25% 1,20% IF, % 1,15% 1,10% 1,15% 1,10% 1,05% 1,05% 1,00% Year 1,00% 42 In order to preserve consistency and comparability of numbers, the average weighted interchange fee has been calculated only for countries with a complete series of observations over the period and also a full set of observations for both the MasterCard and Visa networks. Therefore, the average values presented in this graph may somewhat diverge from the average values calculated from the individual country levels in Graph 7 (year 2004 only), as the latter may include more country observations than were used for the calculation of the average levels in Graph 6. 24

46 b) Member State averages The inquiry revealed significant variations in the weighted average of domestic interchange fees across the EU-25 Member States (see Graph 7 43 ). The difference between the highest (above 1.5%) and the lowest weighted average fees for credit cards in 2004 was about 250% 44. Graph 7 Weighted Average Domestic Interchange Fees Charged to Acquirers on MasterCard and Visa Credit Cards, %, ,50% 2,00% MC Visa Domestic IF, % 1,50% 1,00% 0,50% 0,00% A B C D E F G H I J K L M N O P Q R S T U V W Member States A separate time series analysis ( ) revealed no significant variation in the weighted average domestic interchange fee in the majority of EU-25 countries in both the MasterCard and Visa networks. Germany, Spain, Ireland, Slovakia and the UK were among the countries with a moderate variation in the respective weighted average interchange fee levels. In Italy and Sweden, on the other hand, quite significant changes in the weighted domestic interchange fees were reported only for the MasterCard network and only for In the majority of countries, however, the variation in domestic interchange fees between the MasterCard and Visa networks was quite limited, with only four countries (country 1, 2, 3, 4) having substantially higher interchange rates for MasterCard. Excluding these countries, the average variance between domestic interchange fees charged on credit cards in the MasterCard and Visa networks amounts to 6%. A series of cross-checks was performed on the four outliers above with significantly differing weighted average domestic interchange fees in the MasterCard and Visa networks. The objective was to look for a persistent pattern between discrepancies in the fee levels and differences in the market presence 45 of the two networks in question. The result of these checks revealed that in two of these countries (1 and 2), where the weighted average domestic fee was significantly higher in the MasterCard network than in the Visa network, MasterCard had a 43 The numbers given in this graph do not necessarily correspond to the officially announced domestic interchange fee level in a given country due to the existence of bilaterally agreed on-us fees. Thus, the levels depicted in the graph represent the weighted average levels of all fees applicable in a country (including the on-us fees). 44 For some countries, the level of interchange fees is reported only for one network. However, this does not imply that the other network is not active in the respective geographical market. 45 The data on market shares were calculated based on figures from the RBR Reports on Payment Cards, Western Europe 2006 and Central and Eastern Europe

47 much weaker market presence, measured in terms of the numbers of cards issued in the two markets. In country 1, MasterCard had about 4 times fewer cards issued than Visa, which is equivalent to 20% of the joint MasterCard-Visa issuing market 46. In country 2, MasterCard again had a minority of less than 40% of the joint MasterCard-Visa market in terms of cards issued. In country 3, on the other hand, the substantial difference seen between the interchange fees charged in the MasterCard and Visa networks seems not to correspond to the above logic. In fact, the MasterCard network not only has a much higher weighted average interchange fee, it also has more than 60% of all cards issued on the joint MasterCard-Visa issuing market. Finally, similar checks were carried out on the acquiring side, by looking at market shares based on the number of outlets accepting cards, to see whether a lower interchange fee has been used by networks to increase the market penetration rate 47. The results were inconclusive, however (i.e. no single pattern has been identified). In fact, in 95% of Member States, the MasterCard and Visa brands had almost equal acceptance (about 50% share of the joint network), which implied that most outlets (i.e. merchants) had opted for joint MasterCard- Visa acceptance instead of choosing a single network. The pattern was identical for countries with equal and very different individual MasterCard and Visa shares in terms of cards issued. The only exception was country 4, where MasterCard acceptance in outlets was almost three times higher than for Visa (in 2004: 75% versus 25%). At the same time, MasterCard had on average a higher interchange fee than Visa in Slovenia, although the difference was not substantial. 3.3 Level of interchange fees: domestic interchange fees for debit cards This section surveys interchange fee levels in domestic debit card systems Not all domestic systems have interchange fees It should first be noted that POS interchange fee agreements between banks in open payment card systems are not an intrinsic feature of these systems. The table below shows the EU countries where banks cooperate in payment card systems without charging one another interchange fees for POS transactions. Table 2 FIN LX DK NL Name Pankkikortti Bancomat Dankort PIN 46 The joint MasterCard-Visa issuing market does not include any market share held by other payment networks, such as Diners Club, Amex and JCB. Thus, the calculated market shares of MasterCard and Visa always add up to 100%, irrespective of the presence of other networks in the market. This approach is adopted because of networkspecific characteristics, which imply that four-party networks such as MasterCard and Visa are intrinsically different from three-party networks (Diners Club, Amex and JCB) and therefore need to be analysed separately. While MasterCard and Visa clearly take notice of the pricing and market shares of the three-party networks, they still mostly compete with each other. Accordingly, the analysis often refers to the artificial joint MasterCard-Visa market. 47 The data on the number of outlets accepting cards in each corresponding network were calculated based on figures from the RBR Reports on Payment Cards, Western Europe 2006 and Central and Eastern Europe

48 3.3.2 Comparison of weighted average fees (MC/Visa) a) EU-wide comparisons The analysis of the weighted average domestic interchange fee for debit card transactions across the EU (see Graph 8) reveals that up to 2002 Visa seemed to have on average a higher interchange fee for debit cards than MasterCard. As of 2002, however, the weighted domestic interchange fee on Maestro transactions started to exceed the fee for Visa debit, the average difference over the following three years being about 11%. Interestingly, there were almost no changes in either the Maestro or the Visa debit weighted average domestic fee levels up to 2002, coincidentally the year of the Visa Decision, when suddenly the Visa debit average interchange fee fell sharply, thus leading to a difference of more than 13% between the weighted average debit interchange fees in the two networks. In contrast, the Maestro weighted average interchange fee started falling only as of 2003 and at a much more moderate rate. As the graph shows, the general trend in both networks appears to be a falling one. Having said this, it needs to be added that over the whole period Visa has seen a much more pronounced fall in the weighted average domestic debit interchange fee than MasterCard (16% vs 5%). Graph 8 1,000% 0,950% Weighted Average Domestic Interchange Fee - Visa Electron vs. Maestro, , % MC Visa 1,000% 0,950% 0,900% 0,900% IF, % 0,850% 0,800% 0,850% 0,800% 0,750% 0,750% 0,700% 0,700% 0,650% Year 0,650% Domestic debit card transactions account for the largest and economically most significant share of card payments in Europe (see Graph 9). Historically, debit cards were often the first type of payment card to be introduced and hence customised debit card networks have evolved in a number of Member States. These networks still carry the majority of card transactions in many countries. In addition, MasterCard and Visa also offer domestic debit cards. These products do not have large market shares in all countries, but have been introduced as standard debit cards mainly in the new Member States of Eastern Europe, which had historically not built up national card networks. 48 In order to ensure consistency and comparability of numbers, the average weighted interchange fee has been calculated only for countries with a complete series of observations over the period and with a full set of observations for both Maestro and Visa debit. Therefore, the average values presented in this graph may somewhat diverge from the average values calculated from the individual country levels in Graph 10 (year 2004 only), as the latter may include more country observations than were used for the calculation of the average levels in Graph 8. 27

49 Graph 9 Proportion of Number of Debit Card Transactions Vs Number of Credit Card Transactions in New Member States, 2004, % Proportion of Number of Debit Card Transactions Vs Number of Credit Card Transactions in Old Member States, 2004, % Credit ; 21% Credit ; 35% Debit; 65% Debit; 79% Source: RBR Report, 2006 (based on overall number of transaction volume). In examining domestic debit card payments, this inquiry therefore looks at traditional domestic debit card systems as well as MasterCard and Visa debit card products. b) Member State averages It should first be noted that in some countries where Maestro and Visa debit branded cards are issued, these cards may not be relevant for domestic payments. Thus, domestic interchange fees may not be set for such cards. In some old Member States 49 with established domestic debit systems, domestic debit cards are co-branded with an international debit card logo (e.g. Maestro or Visa debit) to allow mostly for cross-border operability. Furthermore, it is often the case that when a domestic transaction occurs at a local point of sale, it is the domestic debit interchange fee rather than the Maestro or Visa debit domestic interchange fee rate which is paid by an acquirer to an issuer. This is because these transactions are processed under the domestic debit scheme. Where Maestro and Visa debit interchange fee rates are not relevant, these countries are excluded from the analysis. 50 In other countries such as Sweden, there is no domestic debit card network, so Maestro and/or Visa debit interchange fees are of particular importance to the overall cost of card acquiring in the country. Similarly, a number of new Member States (e.g. Czech Republic, Estonia, and Latvia) have no national domestic debit network either, while in Hungary the share of cards issued under a domestic card network logo is negligible. For this reason, Maestro and/or Visa debit remain de facto the most widely used debit card brands in these countries. The results of the inquiry revealed significant variation between weighted average domestic interchange fees on debit cards in the two major international networks, i.e. MasterCard and Visa. The data for the weighted average interchange fees on domestic debit transactions in the two largest international networks are summarised in Graph The fees are ranked according to the maximum value in any of the two networks. 49 For instance in the Netherlands (PIN domestic network) or Belgium (Bancontact domestic network). 50 To provide a full picture, the analysis includes countries with insignificant (though non-zero) Maestro and Visa debit card shares, for example Italy and Denmark (less than 10% in terms of cards issued). 51 The numbers reported in this graph do not necessarily correspond to the officially announced domestic interchange fee level in a given country due to existence of bilaterally agreed on-us fees. Thus, the levels depicted in the graph represent the weighted average levels of all fees applicable in a country (including the on-us fees). 28

50 Graph 10 Domestic IF, % 1,80% 1,60% 1,40% 1,20% 1,00% 0,80% 0,60% 0,40% 0,20% 0,00% Weighted Average Domestic Interchange Fee on Maestro and Visa Debit Cards, %, 2004 Maestro Visa Debit A B C D E F G H I J K L M N Member States The highest weighted average interchange fees on debit card transactions are observed in some of the new Member States. The difference between the highest and lowest weighted average domestic interchange fees on Maestro cards was 1.3 percentage points (around 1300% absolute difference), compared to 1.5 percentage points on Visa debit cards (about 800% absolute difference). If the lowest interchange fee value is excluded from the sample to avoid possible bias, the absolute difference becomes 300% in Maestro and 400% in Visa debit. As with credit cards, debit card interchange fees are characterised by a somewhat low degree of inter-network variation. In fact, according to the available data, only in two countries were the fees for Visa debit cards significantly higher than the fees for Maestro, while in one country the situation was reversed Comparison of fee levels in domestic card networks (debit and credit cards) While most credit card transactions in the EU are transactions with MasterCard or Visa branded payment cards, the bulk of domestic debit transactions run on domestic debit networks. Most domestic card networks offer debit cards only, with a few exceptions. These networks are country-specific, i.e. operate only within a single country, and for the most part lack interoperability among each other. In many countries, the networks were historically run by a consortium or an association of local banks, which sometimes jointly owned the network. The table below lists the major domestic payment cards issued in the EU-25 and the scheme owners. Table 3 Main domestic POS payment cards (2004) Country Card brand Scheme owner BE Bancontact/Mr. Cash Banksys DE EC Cash Zentraler Kreditausschuss FI Pankkikortti Finnish Bankers Association DK Dankort Dankort AS FR Carte Bleue Groupement de Cartes Bancaires NL PIN Currence HU GiroBancard GiroBancard Ltd. (OTP Bank) IT Pagobancomat Cogeban IT Moneta Setefi Spa LU Bancomat Groupement Bancomat PT Multibanco SIBS 29

51 SI Karanta Ljubljanska bank, SKB bank and Abanka SI Activa Banka koper, Nova KBM and Banka Celje SI BA N/A IE Laser Laser LV Unikarte N/A The interchange fee patterns in these domestic debit networks are quite different and diverse, both in terms of fee structure and level of fees. Some systems use flat-rate interchange fees while others use a percentage and some use a combination of both. However, in contrast to MasterCard and Visa, most national debit card systems do not use different interchange fee tiers distinguishing between types of card or types of transaction. A comparison of nominal flat-rate fees, ad valorem fees and combined (i.e.: ad valorem and flat rate) fees necessitates a simulation. Two simulations are provided below for a small ( 5) and a medium-sized ( 50) transaction value. The value of 50 is a good proxy for an average debit card transaction in Europe. Graph 11 POS cards in EU 25: Interchange paid for 50 transaction 0,70 0,60 0,50 0,40 0,30 0,20 0,10 0, V electron Maestro When analysing the above graph, it is important to note that card brands 10, 11 and 13 have comparatively low domestic volumes. The other domestic payment cards 1 to 9 and 12 are the most important debit card in their respective countries. It appears that the level of interchange fees for eight domestic debit cards in the EU-25 is below those for Maestro and Visa debit cards while four domestic debit cards are above (though three of these four cards are unimportant in terms of transaction volumes). Similar results emerge when a 5 transaction is used as the basis for comparison. However, it is notable that for small transactions payment cards with a flat fee are more expensive than cards with an ad valorem fee. 30

52 Preferential interchange fees in domestic card networks A specific issue relevant for competition within the MasterCard and Visa systems is the co-existence of bilaterally and multilaterally agreed interchange fees. The former are often referred to in the industry as on-us fees. Strictly speaking, on-us transactions are transactions where one bank is both the issuer and the acquirer. However, in countries where an inter-bank association acquires, for instance, Maestro or MasterCard transactions, local banks that are co-shareholders of this inter-bank association may be able to offer lower fees to the association. This means that parties to these on us agreements can offer lower merchant fees and thereby prevent new competitors form entering a market. An informal complainant alleged that that the structure and level of domestic interchange fees in the Visa system in Portugal were discriminating against foreign acquirers. The incumbent acquirer, UNICRE, could agree preferential on us interchange fees with domestic issuing banks while foreign acquirers could not obtain equally low fees and had to pay the higher fallback rates. UNICRE is co-owned by Portuguese issuing banks. In Belgium, similar allegations have been raised against Bank Card Company, a joint venture of Belgian banks. It is claimed that after Citigroup entered the market, local banks and Bank Card Company agreed preferential rates while continuing to charge the higher fallback rate to Citigroup. An informal complainant furthermore alleged that in Austria domestic issuers have agreed with the incumbent acquirer, Europay Austria, to set low interchange fees in specific cases for MasterCard (e.g.: petrol station interchange fees) and Maestro transactions (e.g.: food and retail interchange fees). Similar arrangements apply between Visa Austria and its shareholder banks. Europay Austria as well as Visa Austria are co-owned by Austrian issuing banks. The complainant pointed out that this co-existence of preferential ( on us ) interchange fees and general fallback interchange fees created market entry barriers for foreign banks. On us rates also appear to be used in Spain. Industry reports mentioned Spain as difficult to enter for central acquirers as they have to compete with banks that benefit from low on us interchange fee agreements. France would be difficult to enter for central acquirers for the same reason. 4. Conclusion and analysis Interchange Fees the facts The divergence of interchange fees in national card payment systems is considerable. The structure of national interchange fees is also very heterogeneous, as some systems set flat interchange rates while others charge a percentage per card transaction or a combination of flat rates and a percentage. Turning to Visa and MasterCard, the level of domestic interchange fees diverges considerably from one EU Member State to another, even though country-specific differences in these systems are less pronounced in relative terms than in the national card payment systems. In 2004, the nominal rates diverged more than 200% for debit cards and more than 300% for credit cards, and weighted averages diverged up to 250% across the EU for credit cards and up to 400% for debit cards. Within a single country, the relative difference between MasterCard and Visa fees is typically very limited, with the exception of three Member States. As to the structure of domestic interchange fees, it is interesting to note that both MasterCard and Visa have different levels of domestic interchange fees according to merchant segments. Visa banks have set merchant-specific rates in 24 out of 25 EU Member States and MasterCard banks in 10 31

53 out of 25. Most commonly, specific interchange fees are to be found in the petrol and airline sectors. In the same merchant segment and the same card system, the nominal level of domestic interchange fees diverges as much as 165% across the EU. The structure of domestic interchange fees also varies from one country to another. A transaction with the same card type and brand may be e.g. a flat rate in country A and a percentage in country B or a combination of both in country C. Visa and MasterCard sector-specific interchange fees for domestic payments typically diverge to some extent, with MasterCard fees typically above Visa s. Interchange fees for cross-border payments with international debit cards diverge with regard to their structure. It is interesting to observe that one of the international systems uses a flat fee for debit cards while the other system uses a percentage. This has implications for the overall amount of interchange fees paid per card transaction, as a flat fee will yield higher revenues where transaction values are low while the opposite will be true if transaction values are high. Thus, if average transaction values with debit cards drop in the long run, Visa issuers would obtain higher revenues than MasterCard issuers due to the structure of their cross-border fees. As to cross-border interchange fees, the fees for debit card transactions in the Visa system slightly exceeded those of MasterCard until 2001, when this trend reversed. Between 2001 and 2004, MasterCard weighted average cross-border interchange fees remained stable and significantly above the Visa cross-border interchange rate, which fell in the same time period. This trend is even more pronounced with cross-border interchange fees for credit cards. Interchange Fees analysis In a POS system, agreements on interchange fees lead to a transfer of revenues from acquirers to issuers and thereby distort price competition between acquiring banks. Interchange fees also have an effect similar to a tax on each payment with a card at a merchant outlet. The Commission has in the past considered that multilaterally set interchange fees in the Visa system restrict competition between banks for providing services to cardholders and to merchants, as they largely determine the fees charged to both consumer groups. Visa interchange fees were allowed only after Visa committed itself to set interchange fees on the basis of objective costs incurred by issuers for providing concrete services to merchants and to allow member banks to disclose these fees to merchants (cf. the Commission s Visa Decision of 24 July 2002, OJ L 318/17 of ). In subsequent years, national competition authorities such as the UK Office of Fair Trading, the Spanish Tribunal for the Defence of Competition and the Italian Central Bank have concluded that interchange fee agreements infringe competition law, but that they could be allowed if the fees were set on the basis of costs incurred by issuing banks for providing card-related services. The Commission s sector inquiry provides indications that interchange fees are not intrinsic to the operation of card payment systems, as several national systems operate without an interchange fee mechanism. The use of interchange fees may, however, serve several purposes. From a competition viewpoint, it would appear important to what extent interchange fees are de facto (also) used as tool to extract rents from merchants. In this context, some of the preliminary findings in this report, in particular those showing strong country divergences in interchange fees and between merchant segments, may provide indications that the setting of interchange fees could possibly be a matter of market power in some EU Member States. Moreover, there are indications that the setting of interchange fees in the international systems may possibly have the object and/or effect of creating market entry barriers to competition between local and foreign member banks. Both MasterCard and Visa allow the parallel existence of multilaterally set ( fallback ) and bilaterally set ( on us ) interchange fees. While multilateral fees apply to all domestic payments in a given country (irrespective of the bank s 32

54 identity), bilaterally agreed fees only apply between the parties to the agreement. Therefore, in countries where local banks wish to set low interchange fees specifically for certain merchant segments (e.g.: food retail sector, petrol sector), they have a basic choice. They can either set these rates by multilateral decision in a local board or they can go through the more burdensome route of setting the same rates in several bilateral agreements between each issuer and each acquirer in a given country. Under the network rules of MasterCard and Visa, only in EU Member States where local banks set merchant-specific rates multilaterally in a local board are foreign banks able to benefit from such preferential rates. If, on the contrary, the same rates are set in a bundle of identical bilateral interchange fee agreements, the foreign bank pays higher fallback rates. A comparison of the absolute levels of MasterCard and Visa domestic interchange fees suggests that the relatively high level of some merchant-specific rates as opposed to others may have historical reasons and/or may be a question of market power. Turning to the analysis of cross-border interchange fees, the evolution of MasterCard and Visa fees between 2001 and 2004 raises the question why the weighted average of MasterCard crossborder interchange fees for credit cards increased from 2002 even though Visa s weighted average interchange fees for cross-border payments decreased from that year onwards. In other words, does inter-system competition between MasterCard and Visa act as a disciplining market force on bodies setting interchange fees in these networks? The development of MasterCard cross-border interchange fees would rather suggest that inter-system competition did not restrain MasterCard from maintaining higher cross-border interchange fees than those of Visa over more than three years (2002 to 2004). Market forces may therefore be insufficient to penalise card systems with relatively high interchange fees, at least as far as fees for cross-border payments are concerned. 33

55 V. Merchant Charges In this chapter we will analyse the fees paid by merchants for accepting payment cards ( merchant service charges or merchant discount rates ). We will set out the commercial factors driving banks in setting merchant service charges, the way these fees are negotiated with merchants and the extent of blending : a business practice of charging the same fees for different card brands. We will then benchmark average fee levels charged to small versus large merchants, both for credit and debit cards and in both a static and a dynamic analysis. We will furthermore analyse to what extent merchant service charges are lower in countries where regulators or national competition authorities have intervened in the setting of these fees. 1. Structure of merchant service charges 1.1 Description of merchant service charges A merchant service charge (MSC) is the price that a merchant has to pay per transaction to the acquirer, which processes the merchant s transaction through the network and obtains the funds from the cardholder s bank (issuing bank). The transaction is considered to be executed when the transaction amount is debited from the consumer s account and, after deduction of the MSC, is credited to the merchant s account. Most of the cost of using the card is believed to be borne by the merchants 52 and thus, ultimately, by the consumers. There is rarely a difference between the price paid by consumers who use payment cards and that paid by other consumers who use alternative means of payment, such as cash (i.e. the rare practice of surcharging). Indeed, certain payment card networks have explicit rules which prohibit surcharging consumers who use payment cards (see Chapter XIII, section 2). More fundamentally, empirical studies 53 suggest that in practice merchants are reluctant to surcharge customers who use cards, even where this is permitted 54. By charging the merchant, an acquiring bank is believed to pass on the cost of the interchange fee it pays to an issuing bank. It is widely believed that the interchange fee accounts for a substantial part of the merchant service charge. The remaining part is believed to cover other acquiring costs as well as a profit margin. Some acquiring costs are not normally included in the MSC fee. For example, the majority of acquirers stated that, when leasing a terminal to the merchant, the charge for the lease typically does not constitute part of the MSC fee. A few acquirers claimed to charge terminal fees as a component of the MSC. Different acquiring institutions may have different business or pricing policies. For example, some acquirers claim that they do not extract significant (if any) profit from their acquiring business and offer it as a supplementary service to existing clients. Others, on the contrary, see acquiring as a profitable activity and by running it together with issuing enjoy synergies and substantial profits (see Chapter VIII for a detailed profitability analysis of issuing 52 For instance Guibourg and Segendorf, Do Prices Reflect Costs? A study of the price- and cost structure of retail payment services in the Swedish banking sector 2002, Sveriges Riksbank Working Paper Series No 172 October Cf market studies for the Commission ( and by the Office of Fair Trading, UK Payment Systems, An OFT market study of clearing systems and review of plastic card networks of May One of the reasons is that merchants fear their competitors will not start surcharging, which in turn might lead to a loss of customers and turnover. In other words, merchants face a prisoner s dilemma, because they have to assess the potential reaction of their competitors. If all merchants start to surcharge, no loss of turnover has to be feared. However, no merchant can predict that his competitors will start surcharging. This deters merchants from surcharging even when it is permitted by a card system s rules. 54 Ibid. 34

56 and acquiring). All this inevitably affects the way the MSC is negotiated between acquirers and merchants. However, it should not be taken for granted that the MSC is always a negotiable fee. Across the EU-25, various acquirers tend to opt for different pricing methods, sometimes, for example, negotiating individually only with bigger clients. 1.2 Merchant service charges for different categories of merchants Merchant service charges: small vs large merchants Anecdotal evidence based on statements from some market players points to different prices for merchants of differing size. This evidence suggests that smaller merchants, i.e. SMEs, tend to get less favourable deals in terms of merchant service charges than larger merchants. This section, therefore, looks at the empirical evidence derived from the replies of the responding acquirers. The following analysis is subject to certain limitations as it assesses solely the size parameter of a merchant. It does not take into account other merchant-specific characteristics such as, for example, the better risk rating associated typically with larger merchants, which, where of relevance for the analysis, may lead to somewhat biased results. This analysis is not a substitute for a more detailed econometric analysis of the factors affecting the level of MSC, conducted in a later section of this report, but is meant rather to provide an idea and indication of the level of the MSC for different types of merchants. Furthermore, the very fact of having different pricing for smaller merchants versus larger merchants may not constitute discriminatory treatment in itself. It may be justified to some extent by the fact that larger merchants bring higher transaction volumes and therefore may significantly scale down the fixed costs. The lower price may therefore reflect the lower costs incurred with larger merchants. What this analysis will be able to determine with confidence is (A) whether the prices for different sizes of merchants differ, leaving scope for either a possible discriminatory interpretation or for the argument that acquiring is a scale business, and (B) how the relative difference in pricing evolves over time across different payment card networks. a) Credit cards As the replies of the respondents indicate (see Graph 12), in 2004 all payment card networks charged on average much higher MSC rates for credit cards to smaller merchants (as compared to larger merchants). Whereas absolute MSC levels were higher in the Amex, Diners Club and JCB networks, the relative difference in MSC rates between the two groups of merchants ( smaller and larger merchants) was considerably higher in the MasterCard and Visa networks (around 70% in both networks). If this difference is solely explained by the scale factor, it would imply that larger merchants incur half the fixed costs of smaller merchants 55. Interestingly, in the Amex, Diners Club and JCB networks this relative difference in MSC levels was almost half that for MasterCard and Visa (35-50%) It needs to be noted, however, that large merchants may also coincidentally represent sectors that may benefit from lower interchange fees, such as petrol and food retail chains, which consequently may explain the lower merchant service charges. For the analysis to be precise, the comparison of small and large merchants should be done within the same sector, i.e. for the same level of interchange. In the present analysis, however, due to data limitations, this factor has not been controlled for. 56 This, among other things, may also be explained by the fact that, for instance, the Amex card differs significantly from the MasterCard or Visa card, and even more so in terms of market segment (Amex seems to offer mostly corporate cards to large international companies with travelling staff, e.g. T&E companies, while MasterCard and Visa have a much wider market coverage). Amex acquirers may therefore have market power when negotiating 35

57 Graph 12 Comparison of Weighted Average MSC levels Charged on Credit Cards to Larger and Smaller Merchants across EU, 2004, % MSC, % 4,00 3,00 2,00 +70% ~+70% ~+50% +35% ~+40 % Larger Merchants Smaller Merchants 1,00 0,00 MasterCard Visa Am erican Express Payment Card Networks Diners Club JCB As regards the evolution of the difference in the level of MSC between the two groups of merchants, it can be noted that the absolute difference in MSC level (see Graph 13) fell significantly in both the MasterCard and Visa networks over the period (with the exception of a rise for MasterCard in 2001). However, since this difference is absolute rather than relative, and given the drop in the overall MSC level in both networks over the period examined (see section 3 of this chapter), it may give a distorted picture (the absolute level of the MSC may be getting smaller along with the difference, thus providing no measurement of possible discrimination between the two groups of merchants). Therefore, instead of focusing on absolute differences, the analysis needs to look at the relative differences in order take into account possible changes in absolute MSC levels. with merchants in a particular segment and may offer them less profitable deals than those offered by MasterCard and Visa acquirers. This may therefore result in relatively higher MSC rates charged to such merchants for Amex cards and therefore a lower price difference between small and large merchants in this network. 36

58 Graph 13 1,20 Evolution of Absolute Difference in Level of MSC charged on Master Cards and Visa Credit Cards between Larger and Smaller Merchantsacross EU, , % 1,00 Difference in MSC, % 0,80 0,60 0,40 0,20 Difference (MC) Difference (Visa) 0, Payment Card Networks As predicted (see Graph 14), unlike the absolute differences in MSC levels between smaller and larger merchants, the relative differences did not change significantly over On the contrary, and particularly in the Visa network, the difference between MSC levels charged to smaller and larger merchants stayed fairly flat throughout the whole period. In 2004 as compared to 2000, the relative MSC difference for Visa increased from 65% to 70%, while the relative difference for MasterCard slightly fell from 73% to 72%, suggesting that both the main international networks seem to be giving larger merchants similar relative discounts on MSC as compared to smaller merchants. Graph 14 0,80 Evolution of Relative Difference in Level of MSC charged on Master Cards and Visa Credit Cards between Larger and Smaller Merchantsacross EU, , % Difference in MSC, % 0,70 0,60 0,50 0,40 0,30 0,20 0,10 Difference (MC) Difference (Visa) 0, Payment Card Networks 37

59 b) Debit Cards When, however, the MSC levels charged to smaller and larger merchants for debit card transactions are analysed, it can be observed that smaller merchants receive much less favourable treatment from the largest international networks (MasterCard and Visa) in terms of fee rates. In the national debit card networks, which carry the bulk of payment card volumes (national debit transactions sometimes account for about 80% of the total domestic card transaction volume in a given country), smaller merchants on average have to pay only 7% higher MSC rates as compared to larger merchants. The difference in the fees charged to small and large merchants is about 10 times smaller in national debit networks than in the MasterCard and Visa networks (both for debit and credit cards) 57. If the explanation for the different treatment of larger versus smaller merchants is only the scale factor, it should also apply to the national debit networks. As the difference in national debit networks is much smaller, however, it may be asked whether small merchants pay a premium for the use of the MasterCard and Visa networks and what justification there is for this. This question could usefully be studied in more detail. Graph 15 Comparison of Weighted Average MSC levels Charged on Debit Cards to Larger and Smaller Merchants across EU, 2004, % 2,0000 ~+70% ~+60% MSC, % 1,5000 1, % Larger Merchants Smaller Merchants 0,5000 0,0000 Maestro Visa Debit National Debit Payment Card Networks The evolution of absolute differences in MSC charged to different groups of merchants reveals that the difference in absolute terms is falling for both Maestro and Visa debit cards, while it is rising for national debit cards. As previously, an analysis of relative differences is 57 These figures for the MSC in national (domestic) networks, however, do not cover other per-transaction fees that merchants in some countries may end up paying to local processors. These per-transaction processing fees may differ for small and large merchants, given their different transaction volumes and hence different scale factors. MasterCard and Visa MSC rates, however, seem to be final prices paid per transaction and therefore already include processing fees. Thus, once adjusted for the possible supplementary per-transaction processing fee, the price difference for small and large merchants in national networks may be somewhat greater. Where the processing fee is already incorporated in the final MSC paid in national networks, the somewhat lower price difference for small and large merchants may also be explained by the fact that national processors tend to handle the entire volume of payments, including credit transfers and direct debits. This significantly raises the overall scale of the processing and therefore may, to some extent, weaken the incentive of processors (and therefore acquirers if the fee is passed on through acquirers) to reward larger merchants for higher volumes. This therefore may, among other things, limit the price difference for larger and smaller merchants. 38

60 needed in order to determine, irrespective of the trend in the absolute MSC levels, whether the difference did actually change across time. Graph 16 1,00 Evolution of Absolute Difference in Level of MSC charged on Maestro, Visa Debit and National Debit Cards between Larger and Smaller Merchants across EU, , % Difference in MSC, % 0,75 0,50 0,25 Difference (Maestro) Difference (Visa Debit) Difference (Nat.Debit) 0, Payment Card Networks As with the relative differences for credit cards, the relative differences in MSC levels charged to small and large merchants for debit card transactions stayed rather flat over time, i.e. with no significant fluctuations (see Graph 17). As noted above, while these differences are quite similar in the Maestro and Visa debit international networks, they are significantly lower in national debit networks (about 10 times lower in 2004). 39

61 Graph 17 Evolution of Relative Difference in Level of MSC charged on Maestro, Visa Debit and National Debit Cards between Larger and Smaller Merchants across EU, , % 0,70 0,60 Difference in MSC, % 0,50 0,40 0,30 0,20 0,10 Difference (Maestro) Difference (Visa Debit) Difference (Nat.Debit) 0, Payment Card Networks Merchant service charges: various merchant sectors The data obtained from respondents indicate that there is great variation in the MSC levels charged across different sectors in the EU-25 (see Graph 18). The total weighted average MSC range across all networks and across all merchant sectors amounts to 2.5 percentage points, whereas the individual ranges in each separate network (MasterCard, Visa and National Debit) are within 2-3 percentage points. Table 4 Max, Min and Range Weighted Average Values of MSC across Merchant Sectors and per Payment Card Network, 2004, % Master Card Maestro Visa Visa Debit National Debit Max 3,31% 2,45% 2,95% 2,79% 3,00% Min 0,35% 0,43% 0,87% 0,13% 0,29% Range 2,97% 2,01% 2,08% 2,66% 2,71% 40

62 Graph 18 Weighted Average MSC per Type of Merchant Sector across EU, 2004 Florists and Greeting Cards Restaurants Professional Services Memberships, etc. Car Rental Misc. Non T&E Related Manual/ ATM Cash/ Courtesy Check Hotels Misc. Service Providers Airlines Misc. Services Entertainment Medical/Ambulance Misc. T&E Related Merchant Sector Mail Order, Catalogue Order, Telemarktg Misc. and SpecialtyRetail Other Transportation Providers Groceryand Conveninece Stores Auto Maintenance, Repair etc. Auto dealers, Parts and Services Stations Travel Agencies/Services Misc. Repair Services Household Appliances Other Travel High Risk Discount Stores Misc. Business Services Home Furnishings and Improvements Misc. Financial Services Wholesale Trade Fuel Government Services Contracted Services Charitable Organ.,Schools 0 0,5 1 1,5 2 2,5 MSC, % When analysed across all networks, the highest average MSCs are charged in such sectors as Car Rentals, Restaurants, Hotels, Memberships and Service Providers. The lowest average MSCs, on the other hand, are charged for example to Charitable Organisations and Schools, Fuel Companies, Governmental Services and Wholesale Trade Companies. Interestingly, when analysed per network, the results remain mostly the same. In most of the networks, high average MSC rates are charged to Restaurants and Memberships, while low rates are reported for Travel Companies and Wholesale Trade Companies. In contrast, in sectors such as Car Rentals, Auto Dealers, Financial Services and Mail and Catalogue Orders, merchants are charged on average widely varying levels of MSC depending on the network. In order to get a complete picture, a separate sector analysis was performed for larger and smaller merchants, which interestingly revealed a number of discrepancies between the two groups. This analysis covered the two largest international networks (MasterCard and Visa) and 41

63 the national debit networks. While the range in the MSC charged in the various sectors was just slightly higher for larger merchants (2.4 versus 2.3 percentage points), the lists of sectors with the lowest and highest fees looked quite different. Thus, for larger merchants, the lowest average fees (below or close to 1%) were reported in Charitable Organisations, Financial Services, Fuel Companies and Providers of Transport, whereas for smaller merchants the lowest fees were in Contracted Services and Governmental Services. The sectors where the average MSC rates were the highest for larger merchants were Memberships, Hotels, Airlines and Contracted Services, while for smaller merchants the list was headed by Car Rentals, Restaurants, Grocery Stores and Travel Agencies. As can be seen from Graph 19, which shows all the sectors where the average MSC charged to smaller and larger merchants differed by more than 50% in 2004, the relative difference in MSC is more than 80% in some sectors: Auto Dealers, Contracted Services, Financial Services and Grocery Stores,. This may indicate, among other things, that smaller merchants not only in general get less favourable price treatment than larger merchants, but that this price treatment can also differ by merchant sector. Graph 19 Merchant Sectors with the Relative Absolute Difference in MSC Charged to Larger and Smaller Merchants Exceeding 50%, 2004, % 120% Relative Absolute Difference, % 100% 80% 60% 40% 20% 0% Travel Agencies Car Rental Wholesale Trade Medical/ Ambulance Transportation Govmnt Providers Services Merchant Sectors Auto dealers and Services Contracted Services Financial Services Grocery Stores 42

64 1.3 Blended merchant service charges The analysis of the replies from acquirers revealed that blending is a frequent phenomenon across the EU-25. Generally speaking, blending refers to the situation where the same MSC rate is offered to merchants accepting cards issued in two or more different networks. This implies that, facing the same price, the merchant is not inclined to prefer one of these networks, provided that other parameters such as, for example, market share are the same. This in turn can have direct implications for inter-network competition, as price competition essentially ceases to exist. The potential outcome of blending may be higher rates than the merchant needs to pay for acquiring services, given that there is no pressure driving down these charges through inter-network competition 58. Furthermore, blending would tend to nullify the effect of multihoming, which, according to some claims in the academic literature 59, is an important countervailing force to the market power of card payment networks. According to this theory, the very fact that many cardholders simultaneously own payment cards issued by competing networks leads to more intensified competition between the networks, assuming that merchants are sufficiently well informed. Thus, when choosing a network for card acceptance, the merchant, other things being equal (e.g. the degree of acceptance), would opt for the network with the lowest fees. When, however, the merchant faces a single blended rate for card transactions in both (or all) networks, he again becomes indifferent to the choice of network. Inter-network competition arising from multihoming is then cancelled out. As explained previously, this may directly harm merchants as it can lead to higher merchant service charges. Blending seems to be quite widespread across the EU-25. According to respondents, blending usually occurs between MasterCard and Visa card products. In other words, blending occurs between networks with similar levels of interchange fees, and therefore with similar cost components for the MSC. Due to substantial differences in interchange fees, there is no blending between MSC rates charged on debit and credit cards. Furthermore, it seems to be the case that MasterCard and Visa MSC rates are never blended with MSCs for domestic networks, whether credit or debit networks. For the purpose of the analysis, blending by domestic acquirers has been examined separately from blending by foreign subsidiaries 60. The aim here is to see if there is any difference in the pricing models of domestic versus foreign market participants. The data analysis shows (see Graph 20) that almost three in four Member States (72%) have some form of blending of MSC rates. Furthermore, one fifth (Belgium, Denmark, Hungary, Ireland, etc.) seem to have full blending in the market: all of the responding acquirers indicated that they blend MasterCard and Visa MSC rates to 100% of their client base. 58 Blending may also be driven, among other things, by efficiency considerations, if the price difference charged in the two networks is negligible (as this reduces accounting and reporting requirements). On the other hand, there also needs to be some guarantee that once the acquiring cost difference (for example, the interchange fee difference) between transactions in the two networks becomes significant, this is passed on in timely fashion to the merchant in the form of an unblended merchant fee. As blending may create a locked in vicious circle effect, it may therefore still need to be analysed from a competition perspective. 59 Rochet and Tirole (2004, op. cit.), Evans and Schmalensee (2005, op. cit.), see Chapter II on the Economics of the Payment Cards Industry. 60 Foreign subsidiaries, as opposed to purely domestic acquirers, are financial institutions that are involved in domestic acquiring activity but are owned by foreign financial groups or other non-domestic institutions, which have established their commercial presence in a given local market by opening a subsidiary (i.e. foreign subsidiary ). 43

65 Graph 20 Blending across EU, 2004, % Countries w ith full blending among domestic acquiring banks 20% 8% 20% 4% 12% Countries w ith mostly full blending among domestic acquiring banks and full blending among foreign acquiring banks Countries w ith full blending among foreign acquiring banks, and partial blending among domestic acquiring banks Countries w ith partial blending among domestic and foreign banks 36% Countries w ith no blending (neither among domestic, nor among foreign acquiring banks) Countries w ith no information on blending In three member states, Spain, Sweden and Malta, the majority of domestic acquirers and all foreign subsidiaries reported full blending. In Germany, all foreign subsidiaries reported full blending, while the majority of domestic acquirers reported no MSC blending to large merchants (top 25%) and full blending to small and medium-sized merchants (bottom 25%). In nine countries (Czech Republic, Estonia, Greece, Italy, Latvia, Netherlands, Poland, Slovakia and the UK), domestic and foreign acquirers reported, to some extent, partial blending. In some of these countries, particular groups of merchants are treated differently. Thus in the UK, Spain and Italy, some domestic acquirers do not blend MSC rates to large merchants, but blend MSCs for more than 80% of their SME clientele. This may be perceived as a comparative disadvantage for small merchants, who do not benefit from price competition to the same extent as big players. In contrast, Dutch and Slovak acquirers reported that they practiced blending less frequently for smaller merchants than for larger ones. The replies of the respondents reveal that there is for the most part (in 80% of countries) no single policy for blending within a Member State. Thus, in the same country (e.g. Spain, Italy and UK), there can be as many as three different patterns of blending: acquirers who practice no blending at all (A), acquirers that practice partial blending for specific groups of merchants (B), and finally acquirers that offer blended rates to 100% of their client base (C). Interestingly, a separate analysis of large multinational banks with foreign subsidiaries across the EU-25 showed that these banks either have an identical blending pattern across all EU-25 countries (A), or, instead, have quite varying strategies in each country of acquiring (B). In the latter case, they for the most part try to replicate the prevailing blending pattern of domestic banks (going for no blending when domestic banks opt to have none, or instead offering full blending when the majority of domestic banks choose to have this). 44

66 2. Countries with no merchant service charges or highly regulated merchant service charges As stated previously, acquiring institutions are believed to use MSCs mainly in order to pass on to merchants the cost of the interchange fee they pay to issuing institutions for each card transaction. It is also alleged by most market players that interchange fee payments account for the substantial bulk of total acquiring costs. However, in Chapter V on Interchange Fees, it is noted that some domestic debit networks can exist without an interchange fee mechanism while nonetheless securing sufficient transaction volumes and wide card acceptance (e.g: Netherlands, Finland, and Denmark). Accordingly, to ensure an exhaustive MSC analysis, it would be interesting to see whether acquirers in the countries with no interchange fee mechanism still charge an MSC fee to merchants. In this case, this fee does not recoup the interchange fee costs but rather serves as a means to cover the minor residual or other acquiring costs 61 as well as to extract rents from the payment card mechanism. The reported data show that only one of the three EU-25 Member States with no interchange fee mechanism for domestic debit networks in 2004 had no MSC fee, namely Denmark. In the Netherlands and Finland, acquirers did charge merchants an MSC, albeit at a much lower rate than in countries with an interchange fee mechanism for national debit schemes (e.g. Portugal, Italy, etc.). According to the available data, Finland had the lower MSC levels of the two. Denmark, on the other hand, is a Member State with a highly regulated payment card sector. Current national law prohibits acquirers from charging any ad valorem merchant service fee for card acceptance in the domestic debit network (Dankort), so there is therefore an exofficio zero-fee cap. As most of the market players indicate, there was significant pressure from the industry to have this regulation lifted (at least partially) to allow for some MSC. This led to the introduction of a positive MSC in January 2005, which however has since been abolished again (March 2005) Levels of merchant service charges 3.1 Credit cards (MC/Visa) EU-25 comparison As the results of the inquiry show (see Graph ), over the period from 2000 to 2004 weighted average MSC rates on credit card transactions gradually fell across the EU-25 in all international payment card networks, i.e. MasterCard, Visa, Diners Club, American Express and JCB. It is also clear from the results that the level of the MasterCard and Visa weighted average MSC charged on credit cards was much lower than the corresponding level charged in Amex, Diners Club and JCB. Over the whole period, the lowest weighted average MSC fee was charged in the Visa network (average of 1.8%), while the highest was seen with American Express (average of 3.14%). 61 Based on information from market players, other acquiring costs are believed to constitute only a minor portion of acquiring costs. 62 The MSC has been replaced by a fixed annual fee paid by a merchant to an acquirer. 63 The level of the average MSC on MasterCard and Visa credit transactions presented in this graph may somewhat diverge from the average MSC value calculated from individual country levels in Graph 22 (year 2004 only), as the latter may include more country observations than were used for the calculation of the average levels in the time series of Graph 21 (since it included only available observations, while from some countries no complete time series for MSC levels were reported by responding acquirers). 45

67 Over the period, the weighted average MSC rate in the Visa system was on average 9% lower than the corresponding MasterCard rate, the difference growing each year. Interestingly, when trends in weighted average domestic interchange fees charged on MasterCard and Visa and the corresponding weighted average MSC levels are compared, it may be seen that both trends are falling on average. Given that, as noted above, the interchange fee may account for a substantial part of the MSC (for details, see a subsequent section of the report), such similarities in trend may be expected. There is, however, a slight mismatch in trends as far as MasterCard rates are concerned. Thus in 2002, the MasterCard weighted average domestic interchange fee rose by about 2.4%, whereas the corresponding MasterCard MSC level did not increase. This phenomenon has not been further examined, but one possible explanation among others could be that prices are sticky. Graph 21 Weighted Average MSC, % 3,5 3 2,5 2 1,5 1 Weighted Average MSC Rates Charged for Credit Cards by International Payment Card Networks across EU, , % Year MasterCard Visa Diners Club American Express JCB Country analysis The country-specific analysis of the weighted average MSC credit card rates (see Graph 22) revealed a substantial variation in MSC rates across the Member States. In the Visa network, the fees varied from a low of 0.77% up to 3.10%, the difference being around 300%. In the MasterCard network, the lowest weighted average MSC was reported to be 0.95%, while the highest was 2.98%, a difference of more than 200%. Despite some exceptions, most of the countries with relatively high MSC rates for one of the two main international networks had quite high rates for the other one. In general, the MSC levels in the two networks were quite similar within one country for most EU-25 Member States, with an average absolute difference, adjusting for the outlier, of about 11%. In the country outlier, on the other hand, the difference between the weighted average MasterCard and Visa MSC rates was reported to be more than 100%. Not surprisingly, no acquirers in this country reported blending of MasterCard and Visa MSC rates to their merchants. 46

68 Graph 22 3,5 Weighted Average MSC Rates Charged per Country for MasterCard and Visa Credit Cards, 2004, % Weighted Average MSC, % 3 2,5 2 1,5 1 0,5 MasterCard Visa 0 A B C D E F G H I J K L M N Member State 3.2 Debit cards EU-25 comparison As with the trend in the weighted average MSC charged to merchants on credit card transactions, the weighted average MSC levels across all debit card payment networks decreased over the period (see Graph ). The lowest average MSC level was reported in national debit networks, the highest for Maestro cards (MasterCard network) (1.17% vs 1.60%). Across the EU-25 and over the period, the weighted average MSC rate charged in national debit networks was on average 30% lower than the corresponding Visa debit MSC rate, and almost 40% lower than the corresponding Maestro MSC rate. In contrast, the average difference between the weighted average MSC rates charged on Maestro and Visa debit transactions was quite limited and amounted to only about 6%, which is even lower than the corresponding difference between the weighted average MSC rates charged on credit cards by the two networks. This may, among other things, explain the use of blended Maestro/Visa debit and MasterCard/Visa MSC rates in some of the EU-25 Member States. Finally, as with other debit networks, the trend in weighted average domestic debit MSC levels also fell, though to a lesser degree than with the Maestro and Visa debit networks (10% decrease over 5 years as compared to 17% and 20% for Maestro and Visa debit, respectively). 64 The level of the average MSC on Maestro and Visa Debit transactions as presented in this graph may somewhat diverge from the average MSC value calculated from individual country levels in Graph 24 (year 2004 only), as the latter may include more country observations than were used for the calculation of the average levels in the time series of Graph 23 (since it included only available observations, while from some countries no complete time series for MSC levels were reported by responding acquirers). 47

69 Graph 23 2,015 Weighted Average MSC Rates Charged for Debit Cards by International Payment Card Netw orks (Maestro & Visa Debit) and National Debit Networks, , % Weighted Average MSC, % 1,515 1,015 0,515 Maestro Visa Debit National Debit 0, Year Country analysis a) Maestro and Visa Debit As with weighted average MSC levels on credit cards, the weighted average MSC charged on debit cards (see Graph 24) showed considerable variation across the Member States. Here, the analysis looked only at countries where domestic Maestro and Visa debit transactions were possible and/or relevant. Countries with Maestro or Visa-Electron co-branded debit cards, where in fact the transactions are processed in domestic debit networks rather than in either of the two international debit networks, as is the case for example in Belgium and Netherlands, were omitted from the analysis. Moreover, it needs to be noted that due to data limitations only a few countries were included in the analysis. According to the data, the weighted average MSC fees in the Visa network varied from a low of 0.32% up to roughly 1.9%, the difference being around 500%. In the MasterCard network, the lowest weighted average MSC was reported to be 0.36% 65, while the highest was above 2%, which is 6.5 times higher. In general, the variation in weighted average MSC rates charged on debit cards was higher than the variation in MSC levels for credit cards. As in the credit card analysis, the analysis of MSC levels charged on debit cards revealed on average similar MSC levels across the two largest international networks, with one country being an outlier (about 50% difference). Accordingly, while the average absolute difference between Maestro and Visa debit weighted average MSC levels, without adjusting for the outlier, amounted to 13%, with the corresponding adjustment, it came to only about 5%. 65 For some countries, no cross-reference to Visa levels was possible, as the set of countries with available MSC levels for the two networks differed somewhat in each case. Generally speaking, the maxima and minima found in this analysis should be treated as local rather than global, given that no entire set of data was available. Nonetheless, the variation noted will, if anything, be increased by adding countries to the analysis. 48

70 Graph 24 Weighted Average MSC Rates Charged per Country for Maestro and Visa Debit Cards, 2004, % 3,000 Weighted Average MSC, % 2,500 2,000 1,500 1,000 0,500 Maestro Visa Debit 0,000 A B C D E F G H I Member State b) National debit networks While the variation in average MSC fees across the EU-25 between Maestro and Visa debit cards was not pronounced, the MSC fees charged in national debit networks differed significantly from the MSC charged in international networks across the EU-25. As can be seen from Graph 25, which compares weighted average MSC levels on Maestro and national debit cards across a group of EU-25 countries, Maestro MSC rates tended to exceed (in some cases significantly) those for national debit cards (one country being an outlier). For example, in the case of one country, the weighted average MSC rate in the domestic debit network in 2004 was almost half the corresponding rate in the Maestro network. Nonetheless, such a difference cannot be interpreted simply by looking at the absolute values. Instead, a cross-analysis of MSC level versus interchange level within a particular network is needed. For comparison purposes, countries where domestic Maestro is not relevant and/or significant (i.e. Finland, Ireland, Netherlands, Belgium and France) were also included in the analysis in order to demonstrate that the respective weighted average MSC levels are much lower in the corresponding domestic debit networks as compared to the Maestro network. 49

71 Graph 25 Weighted Average MSC Rates Charged per Country for National Debit Cards versus MSC Rates Charged for Maestro, 2004, % MSC, % 2 1,8 1,6 1,4 1,2 1 0,8 0,6 0,4 0,2 0 A B C D E F G H Member States National Debit Maestro 4. Conclusion and analysis Merchant service charges the facts Country discrepancies in merchant fees are strong, both in the national and international systems. Weighted average merchant rates for national payment cards varied between 0.075% and % in 2004, which is a discrepancy of nearly 1500%. Weighted average merchant rates charged for accepting international credit cards varied between 0.77% and 3.10% (approx. 300%) in one system and between 0.95% and 2.98% (200%) in the other. Similarly, strong variances have been reported for international debit cards, with rates ranging from 0.32% to 1.92% (500%) in one system and from 0.36% to 2.75% (650%) in the other. Smaller merchants typically pay higher rates than larger merchants, with discrepancies of 70% for MasterCard and Visa cards, 50% for American Express, 40% for JCB and 35% for Diners Club. 72% of the acquiring banks surveyed charge more or less identical fees to merchants for accepting MasterCard and Visa cards ( blending ). Across the EU-25, merchant discount rates for accepting payment cards vary strongly from one merchant segment to the other. The highest average rates are charged to sectors such as car rentals, restaurants and hotels while the lowest fees are charged to charitable organisations, government services, the fuel sector and the wholesale trade. Merchant service charges - analysis It is interesting for a competition analysis to explore why small merchants on average pay 70% more for payment card acceptance than large merchants. In theory, this could be explained by lower costs for signing up merchants with higher transaction volumes. However, a comparison of price differentials between large and small merchants in the international schemes (MC/Visa: 70%, Amex 50%, JCB 40%, Diners 35%) with those in domestic systems (7% on average) indicates that scale is possibly not the main reason. It could be that smaller merchants pay a premium for accepting MasterCard and Visa cards. If that were true, the 50

72 differentiation of prices according to the size of the merchant could be a measure for the exercise of market power by banks within a given system. It would appear that merchants paying the highest average rates for MasterCard and Visa card acceptance (florists, restaurants, professional services, car rental, hotels) are typically those active in the T&E sector, where travellers expect to pay with cards, while merchants paying lower fees are typically to be found in segments with low profit margins (charitable organisations, contracted services, government services, wholesale trade etc). An outlier is the fuel sector, which yields high margins but nevertheless pays comparatively low fees for card acceptance. 51

73 VI. Cardholder fees 1. Introduction This chapter will provide a descriptive comparison of the fees paid by cardholders for credit and debit cards in the EU-25 Member States over the period The object of this exercise is two-fold. Firstly, it aims to assess to what extent the fees actually paid by cardholders differ across countries and networks. This may in turn permit an assessment of how the prices of relatively comparable services differ across countries. Significant differences might be the result of lack of competition and market integration. Secondly, it aims to test a simple theoretical prediction from mainstream economic theory, namely that increasing the level of the interchange fee would, ceteris paribus, raise merchant service charges but would lower cardholder fees through the interchange fee mechanism. This negative relationship between the level of cardholder fees and the level of interchange fee is also cited by card networks, who claim that without the interchange fees paid by acquirers to issuers, issuers would have to recoup all their costs from cardholders, so cardholder fees would be higher if the interchange fee decreases and lower if it increases. It is thus important to assess how cardholder fees are correlated with the interchange fee and whether any increase in the interchange fee over time is passed on in lower cardholder fees. Cardholders have a contractual relationship with the card issuer. By charging the cardholder, an issuing bank is believed not only to cover the costs of the service provided, e.g. transaction processing and billing, but also to earn a profit margin. However, issuers usually do not charge a single fee but several fees to cardholders. For instance, issuers may charge annual fees such as a fee per card and also transaction fees. Furthermore, issuers may use payment cards as a way to attract costumers to purchase other products, such as current accounts and credit, which may imply that cardholder fees are not determined in a fully autonomous manner. According to an industry expert consulted by DG Competition, debit cards in particular may be an accessory product of current accounts. All these factors imply that different issuing institutions may have different pricing policies and, consequently, a comparison of cardholder fees across countries needs to be undertaken carefully. Accordingly, in order to carry out a meaningful comparison of cardholder fees across countries, respondents were requested to provide data not on a single fee but on a list of fees paid by a typical cardholder, who may hold either a consumer or a business standard/classic card. A typical cardholder is one offered the standard conditions with no special rules or rebates. A standard or classic card is the card most widely issued by issuing institutions, thus excluding the categories of gold, platinum, affinity or any other special cards. The requested information was also broken down by network. Since cardholder fees are not the only source of revenue of issuing institutions, it is important to establish a link between the level of cardholder fees and the network because the level of cardholder fees may be indirectly influenced by the level of interchange fee, as claimed by industry participants. We will compare four relevant cardholder fees. These fees are: (1) the fee per card, which is an annual fee given in euros; (2) the card issuance fee, which is a fee charged only when the card is issued, also given in euros; (3) the fee per transaction, which is charged both as a percentage or in a (euro) amount per transaction and (4) the account statement and billing information fee, which again is an annual fee given in euros 66. These four fees were requested for a typical cardholder with standard/classical credit and debit cards, for each year over the period These fees were selected for our analysis because they contribute 66 Converted into an annual figure. 52

74 significantly to the revenues of the issuing institutions. At the same time, respondents reported more data on these fees than on others such as penalty fees. Among these four fees, the annual fee per card is the most widely used in the EU-25 Member States. In fact, our sample contains more than 600 responses reporting positive values for this fee. This compares with 320 for the issuance fee, 200 for the account statement and billing information fee and 100 for the fee per transaction. The fact that issuers use these fees differently seems to corroborate the idea that pricing is very heterogeneous in the issuing business. 2. Credit cards 2.1 Fee per card Looking at the simple average of the fee per card actually paid in 2004 by a typical cardholder holding a classical or standard credit card in 23 Member States 67, the results show that Diners Club is the network where cardholders pay the highest fee (57 euros). American Express charges on average 47 euros at EU level. Cardholders in the two most important international networks pay relatively the same amount of fee per card annually: 24 euros for MasterCard and 24 euros for Visa. It should be noted that the computed averages for Diners and American Express are based on a rather limited number of responses: 37 for American Express and 17 for Diners Club. In contrast, the computed averages for both MasterCard and Visa are based on a relatively high and equivalent number of observations: 157 for MasterCard and 180 for Visa. We compared the annual average fee across countries in 2004 between the two major networks. Graph 26 shows the country annual average for 23 EU Member States in Graph 26 EURO Country-average fee per credit card, 2004 MasterCard A B C D E F G H I J K L M N O P Q R S T U V W Visa Average fee =24 Euro Member States Analysis of this graph shows that the average fee incurred by a typical cardholder for the two main credit card brands is relatively similar within the countries for which data are available. However, the average fee for both networks varies significantly across countries. 67 Two countries were excluded due to data unavailability. 53

75 Given that the average at EU level for MasterCard and Visa is 23 and 24 euros, respectively, an important number of countries have values clearly above these figures. In contrast, countries U and V are below the average values for the two networks. These figures aggregate the fees charged to a typical consumer and business cardholder holding a standard or classic credit card. However, the fees charged to a typical consumer cardholder may differ significantly from those charged to a typical business cardholder. In fact, the annual volume of transactions generated by each class of cardholder may be significantly different and the general commercial interest that governs the relationship between the issuing institution and its clients (in a broad sense) may also be different. It is therefore interesting to split the data into fees paid by consumer and business cardholders and to assess if these classes of cardholders are charged differently. Graph 27 Country-average fee per credit card by type of card, MasterCard MasterCard-Business MasterCard-Consumer Euro A B C D E F G H I J K L M N O P Q R S T U Member States It can be observed that in 18 of the 21 countries with data on both types of cardholders, a typical business cardholder pays more for a standard/classic credit card than a typical consumer in the MasterCard network. Graph 28 gives the same information for the Visa network. 54

76 Graph Country-average fee per credit card by type of card, Visa Visa-Business Visa-Consumer Euro A B C D E F G H I J K L M N O P Q R S Member States It can be observed that in 17 of the 19 countries with data on both types of cardholders, a typical business cardholder pays more for a standard/classic credit card than a typical consumer in the Visa network. These findings are corroborated by the average card fee at EU level. In fact, a typical business cardholder pays 32 euros for a standard/classic card while a typical consumer cardholder pays 19 euros in the MasterCard network. Similarly, a typical business cardholder pays 34 euros for a standard/classic card while a typical consumer cardholder pays 18 euros in the Visa network. It is also interesting to note that the aggregate fee has evolved differently over time across the countries considered 68 Graph 29 Euro Growth rate fee per card, MasterCard Visa A B C D E F G H I J K L M N O P Q R S T U V W Member States 68 It should be noted that for assessing how the fee per card evolved over the period across countries, only the issuers that provided data over the entire period were considered. 55

77 In some countries, the fee per card remained fairly unchanged from 2000 onwards. Overall, the average rate of growth was about 5% over the period Correlation between fee per card and interchange fee Industry participants and mainstream economic theory suggest that increasing the level of the interchange fee would, ceteris paribus, raise merchant service charges but would lower cardholder fees through the interchange fee mechanism. That is, in the absence of interchange fees paid by acquirers to issuers, issuers would have to recoup all their costs from cardholders, with the result that cardholder fees are higher if the interchange fee decreases and lower if it increases. It is therefore relevant to assess whether there is a strong negative correlation between the average fee per card and the level of interchange fee for a given country and network. In order to establish this correlation, we computed a simple correlation coefficient between the country-average fee per card and the level of the interchange fee for the MasterCard and Visa networks over the period Table 5- Correlation coefficients between fee per card and interchange fee Years Visa MasterCard Results suggest that there is not a strong negative relationship between the level of the cardholder fee and the level of the interchange fee. This pattern is common to both networks and relatively consistent over time. The fact that the low correlation values remained relatively unchanged over time may imply that a possible increase (decrease) in the interchange fee during this period does not seem to have been passed on in lower (higher) cardholder fees. These simple correlation coefficients do not control for other factors that may affect the fee per card level. However, an econometric estimation controlling for other variables that may affect the fee per card level shows that if the interchange fee increases by 1 Euro only 25 cents are passed on to consumers in lower fees 69. This result challenges the hypothesis advanced by some industry participants and the economic literature that an increase in interchange fees exactly equals a decrease in cardholder fees. These results confirm the findings described in the chapter on profitability and may cast doubt on the relevance of the arguments put forward by industry participants and the economic literature as regards the role played by the interchange fee in this industry. 2.2 Card issuance fee The second most widely used fee charged to cardholders in the EU-25 Member States is the card issuance fee 70. An analysis of the data shows that this fee is not applied in 5 Member States. 69 See annex 5 for more details. 70 It should be noted that the issuance fee is not weighted by the validity period of the card. 56

78 In the remaining 20 countries, however, it is interesting to observe that that this fee is, on average, higher for cardholders holding credit cards issued in the MasterCard and Visa networks (14 euros for both) than for those with cards issued by American Express and Diners Club (11 and 5 euros, respectively). Looking at the card issuance fee paid by a typical cardholder for MasterCard and Visa in 2004, it is possible to conclude that there are, on average, significant differences across the 20 countries. Graph 30 Country-average issuance fee per credit card, 2004 Euro MasterCrad Visa A B C D E F G H I J K L M N O P Q R S T Member States It can be ascertained from Graph 30 that certain countries are clearly above the EU average of 14 euros for both networks. Finally, it is important to point out that the country annual average fees for the issuance of credit cards have remained largely unchanged since This also means, as for the fee per card, that any change in the interchange fee does not seem to have been passed on in the cardholder issuance fees. 2.3 Account statement and billing information fee The third annual fee analysed is the fee for account statements and billing information. An analysis of the data shows that, in contrast with the first two fees, no account statement and billing information fee is charged in the majority of Member States. Looking at the respondents reporting positive values, the average fee is relatively similar across the different networks: 0.17 euros for American Express, 0.28 euros for Diners Club, 0.92 euros for MasterCard and 0.64 euros for Visa. Graph 31 shows how much MasterCard and Visa cardholders are charged in the 10 countries where respondents reported positive figures. 57

79 Graph Country-average fee per account statement and biling information for credit card", 2004 MasterCard Visa 5 EURO B C D E F G H I J K Member States The only outstanding result in Graph 31 is observed in country B. In fact, country B is clearly the country that has the highest account statement fee for both MasterCard and Visa cards. Finally, it is also interesting to note that, on average, business and consumer cardholders pay relatively the same fee in these 10 countries. As regards the evolution of this fee over time, it may be noted that the annualised fee for credit cards has remained largely unchanged since Fee per transaction Finally, we have analysed the fee per transaction, defined either in euros or as a percentage. In 19 of the 25 countries, respondents claim that they do not charge their cardholders for each transaction they make, irrespective of the network. For the remaining 6 countries, it is of interest to assess how the fee per transaction varies among them and by network 71. Among those where a fee per transaction is charged as a percentage of the transaction volume, this fee varies from 0.1% to 0.7% in the MasterCard network and from 0.5% to 0.7% in the Visa network. As regards those where the fee is expressed in euros, the limited number of (positive) observations for this fee makes any result difficult to interpret. 3. Debit cards 3.1 Fee per card As with the analysis carried out for credit cards, the fees paid by cardholders for debit cards in the EU Member states are compared here for For the purpose of this analysis, the fees charged for debit cards in the MasterCard, Visa and national debit networks in 20 EU countries are compared. 72. Simple statistics show that, on average, the fee per card is significantly lower for debit than for credit cards. Indeed, the fee per card debit cards is on average 10 euros for MasterCard (Maestro) (as against 24 euros for credit cards) and 11 euros under the Visa brand (Visa Electron) (as against 23 euros for credit cards). This pattern is quite consistent across countries, as demonstrated in Graph 32, which shows the country-average card fee in 2004 for 20 EU countries. 71 Only MasterCard and Visa are analysed, due to data unavailability. 72 It should be noted that for some countries the sample is only for one network. 58

80 Graph 32 EURO Country-average fee per debit card, 2004 Maestro Visa Electron National debit A B C D E F G H I J K L M N O P Q R S T Member States Comparing country-average fees in 2004 for credit and debit cards (i.e. comparing Graph 26 with Graph 32), it can be ascertained that, with the exception of three countries the fee per card is higher for credit than for debit. Graph 32 shows that, as for credit cards, the fees paid in the two international networks are quite similar within countries for which data are available. However, there are important differences in the level of the fees across countries. Another interesting finding emerges from a comparison between the fees in international networks (MasterCard and Visa) and those for national debit networks. The average fee per card in the national schemes amounts to 9 euros (as against 10 and 11 euros for MasterCard and Visa, respectively). On average, the fee per card is significantly lower in national debit schemes than in the international debit networks. 3.2 Card issuance fee An analysis of the data shows that this fee is not applied in most EU Member States. In the remaining 13 countries where this fee is applied, it is interesting to note that cardholders pay an issuance fee of more than 6 euros. 59

81 Graph Country-average issuance fee per debit card, Maestro Visa Electron National Debit Euro A B C D E F G H I J K L M Member States It is also worth noting that the country-average issuance fees for debit cards have remained largely unchanged since Account statement and billing information fee An analysis of the data shows that no fee for account statements and billing information is charged to cardholders in most Member States. Graph 34 shows the country-average fee for debit cards in Graph ,5 3 Country-average fee per account statement and billing information for debit cards, 2004 Maestro Visa Electron National Debit 2,5 Euro 2 1,5 1 0,5 0 A B C D E F G H I J Member States In 8 out of 18 countries where data were available, cardholders are not charged for account statement and billing information. 60

82 3.4 Fee per transaction In 17 of the 25 countries, respondents claimed that they do not charge their cardholders for each transaction they make, irrespective of the network. For the remaining 8 countries, results show that, for those where a transaction fee is charged as a percentage of the transaction volume, the fees vary from 0.1 % to 0.75%. 4. Conclusion This chapter compared four relevant cardholder fees: (1) the fee per card, which is an annual fee given in euros; (2) the card issuance fee, which is a fee charged only when the card is issued, again given in euros; (3) the fee per transaction, which is charged as percentage or as a euro amount per transaction; and (4) the account statement and billing information fee, which is an annual fee also given in euros. These four fees were requested for a typical cardholder with standard/classical credit and debit cards, for each year over the period The annual fee per card seems to be not only the most widely used fee charged to cardholders in the EU-25 Member States but also one of most important elements in the revenue generated by cardholder fees for issuers of both credit and debit cards. The average fee per credit card charged to cardholders in the MasterCard and Visa networks is 23 euros. It is also worth noting that a typical business cardholder pays on average significantly more than a consumer cardholder for a standard/classic credit card in both networks. While the fees charged in these two networks are remarkably similar within a country, they vary significantly across countries. Several countries have fees above the average value of 23 euros for both networks. The average fee for debit cards is 10, 11, and 9 euros for the MasterCard, Visa and national debit networks, respectively. The average credit card issuance fee charged to cardholders in the MasterCard and Visa networks is 14 euros. The average debit card issuance fee is 6 euros. No account statement and billing information fee or fee per transaction is charged in most Member States. The average account statement fee for credit cards is 0.92 euros for MasterCard and 0.64 euros for Visa. The corresponding figures for debit cards are 0.32 euros for Maestro, 0.20 euros for Visa Electron and 0.10 euros for national debit networks. Fees per transaction are only marginal, for both credit and debit cards. All in all, the differences in the fees charged to consumers for card issuing are significant and might be the result of a lack of competition and market integration. Importantly, the analysis reveals no strong negative relationship between the level of the fee per card and the level of interchange fee at either country or network level. The empirical evidence suggests that if the interchange fee increases 1 Euro only 25 cents are passed on in lower cardholder fees. Such result challenges the hypothesis advanced by some industry participants and the economic literature of a perfect inverse relationship between card fees and interchange fees. Finally, all the four fees are on average more than double for credit cards than for debit cards. 61

83 VII. Profitability 1. Introduction The profitability of a business may provide important information for a competition analysis. On one hand, the existence of significant economic profits may be the reward for taking risks and for innovating and/or it may be the reward for superior efficiency and better management. On the other hand, high profits may also be the result of having and exerting market power, in particular if profit margins remain high over a long time period in a relatively mature market. The sector inquiry therefore analysed to what extent issuing and acquiring are profitable and how profits developed during the period from 2000 to The second purpose of our profit analysis was to assess to what extent the profitability of the issuing business depends on revenues generated by interchange fees. In the following chapter we therefore analyse: The profitability of acquiring banks (for the entire EU-25 and per country). The profitability of issuing banks (for the entire EU-25 and per country). The profitability of issuing and acquiring banks analysed together, with special consideration given to interchange fees. The impact of interchange fees on the profitability of payment card issuing. The analysis is set out in two chapters, one for credit and one for debit cards. 2. Profitability trends This chapter provides a descriptive comparison of profitability trends for issuing and acquiring businesses in credit and debit cards for all EU-25 Member States. This analysis covers the period Looking at profitability may yield important information for competition analysis. In fact, while the existence of significant rents may be the reward for taking risks and innovating, superior efficiency or better management, they could also be the result of having and exerting market power. High and persistent rents in relatively mature markets where some prices, such as interchange fees, are determined collectively may suggest the latter. These findings, together with other evidence obtained by this inquiry, may reveal whether a firm or a group of firms is exercising market power to the detriment of consumers in a particular market. Taking advantage of the detailed data set available, this chapter also aims at examining further the role played by the interchange fee in a two-sided industry. As described in the chapter on the economics of the payment card industry, mainstream economic theory claims that the interchange fee keeps both sides of the industry on board and corrects market failures linked to the existence of externalities. Roughly speaking, an optimum level of interchange fee is needed because price mechanisms fail to internalise the existing externalities. The practical consequence is that the issuing business needs to be subsidised by the acquiring business in order to bring sufficient cardholders into the system 73. This type of justification for the existence of interchange fees is also put forward by the payment card systems. For instance, one international network believes that in the absence of the POS interchange fees paid by acquirers to issuers, issuers would have to recoup all of their costs from cardholders and this would lead to a level of card issuing that is not optimal for the system as a whole. In the same 73 Rochet and Tirole, 2002, op. cit., argue that the interchange fee is also needed to compensate for the pricing distortions introduced by the exercise of issuer market power, i.e. subsidising issuers with market power would induce them to reduce their prices. 62

84 vein, another international network argues that interchange fees are a mechanism for redressing the imbalance between issuers and acquirers costs and revenues in delivering a payment card 74. Yet another strand of economic literature suggests that privately optimal interchange fees may be too high, notably if merchant fees increase with interchange fees but issuers do not pass the additional interchange fee revenue back to cardholders. In this case, high interchange fees are a way to transfer rents to the side of the scheme where they are least likely to be competed away (see Wright, 2004, op. cit. and Bergman, 2005, op. cit.). Assessing the extent to which the profitability of issuing business depends on the revenues generated by the current level of interchange fees may provide further insight into the accuracy of the two different theoretical predictions and also the pertinence of the arguments put forward by the payment card systems. The measurement of the profitability of a specific activity is typically subject to problems related to the allocation of costs that are common to other activities. This could be also relevant in our case, because acquirers and issuers (which may be multi-product firms) may also carry out other activities 75. However, it is worth noting that the allocation of revenues and costs, based on accounting data, was made by the respondents. Thus, the measurement of profitability has to be considered reliable because it was made by those who best know their own business. Consequently, the computation of key cost and revenue parameters by the respondents themselves reduces significantly the degree of uncertainty as to their true level. Moreover, the revenues and costs are not separated by the different payment systems in which acquirers and issuers participate, further decreasing this uncertainty. For the purpose of the inquiry, both issuing and acquiring institutions were requested to report their total revenues and total costs associated with the issuing and acquiring of credit and debit cards. The questionnaire provided a breakdown of the most relevant parameters for total revenues and total costs. In the acquiring business, total revenues are given by merchant service charges, terminal processing fees, currency conversion fees, and other type of incomes ; total costs, in turn, include interchange fees, transaction processing costs, and other type of costs. In the issuing business, total revenues are given by interest charged, interchange fees, cardholder fees, currency conversion fees, income from co-branding, and other type of incomes ; total costs include costs for the provision of a free funding period, card production and transaction processing costs, billing, fraud, credit losses, costs related to rebates, staff costs and other type of costs. The parameter other type of income/cost aims to capture any other relevant type of income or cost in the acquiring and issuing of cards, as perceived by the respondents, which does not fall under the other categories. Costs related to the depreciation of assets, for instance, could be included in this category. 2.1 Credit cards Acquiring business a) EU-25 comparison In order to investigate the magnitude and evolution of profitability in the acquiring of credit cards in the EU-25 Member States over the period , we make use of a simple profit-to-cost ratio. The operational profit-to-cost ratio before taxes (hereafter profit ratio ) in the acquiring of credit cards by acquirer B in country A at time t is given by: A Income A B Cost t B t x100. This measure will be used throughout this chapter. A Cost B t 74 See the chapter on interchange fees. 75 According to an industry expert consulted by DG Competition, some institutions may have difficulties in isolating the debit card business from other activities since debit cards may be an accessory product of current accounts. 63

85 Of the 83 respondents that reported figures for the acquiring of credit cards in 2004, 52 reported a positive and 31 a negative profit ratio. Graph 35 shows the distribution of profit ratios in the acquiring of credit cards for 2004 for all respondents (at EU-25 level). Graph 35 - Distribution of profit ratios in the acquiring of credit cards, 2004 Density rate The distribution in Graph 35 reveals that 42 out of 83 acquirers reported profit ratios higher than 8.3% (median value). Of these, 21 acquirers reported figures between 8.3% and 25.7%, while the remaining 21 acquirers reported figures above 25.7%. If one also takes into consideration that the weighted profit ratio of all respondents for 2004 is 15.9%, it seems possible to conclude that the acquiring business for credit cards is relatively profitable at EU-25 level. b) Country analysis However, the analysis of profitability at an aggregated level may mask substantial differences across Member States. Therefore, the analysis of profitability at country level may provide further insight into the distribution of profit ratios across countries. Crucially, competition analysis needs to take into account the situation at the level of geographical markets. In order to avoid giving equal weight to both small and large acquirers in the determination of the overall country profit ratio, this ratio is a weighted average of all the acquirers in the country in question (the weight is given by the total income of acquirers). 64

86 Graph Profit ratio differentials in credit cards acquiring, Profit weighted average ratio = 15,9% Profit ratio (%) A B C D E F G H I J K L M N O P Q R S T U V W Y Z -20 Member States It can be ascertained that the acquiring of credit cards was profitable in 19 of the 25 EU countries in The profit ratios vary from -16% to 62%. It can also be observed that 9 countries are above the EU-25 weighted profit ratio (15.9%). Some of these profit ratios are based on a limited number of observations, which means that results may not be entirely representative of profit ratios for a given country. Additionally, we have no way of assessing whether results are driven by some outliers in the sample. In cases where the sole observations in the sample were for large and specialised acquiring institutions, the problem is less important. In contrast, in countries where only a limited number of non-specialised acquiring institutions provided figures estimated figures, in some cases the results need to be analysed with more caution. For countries with a significant number of responses, we can assess the variability of profit ratios within the country and to what extent the results are driven by outliers in the sample. While the profit ratios of acquirers in some markets are homogenous, respondents in a few countries reported diverging profit ratios. When looking in detail at the data, we discover that differences in the profit ratios within a country are mainly due to differences in their cost structures. These discrepancies arise because some acquirers incur, among other things, significantly higher average transaction processing costs and staff costs than others (see also below for further details on the cost structure of issuing institutions). A static analysis based on a single year may, however, neglect an important issue, which is the stage of the business cycle in each market. It is thus important to address this issue by looking at the way profit ratios evolved during the period in each country. Graph 37 reports the weighted profit ratios by country over the period

87 Graph Profit ratios in credit card acquiring, Profit ratio (%) A B C D E F G H I J K L M N O P Q R S T U V W Y Z Member States The analysis of this graph shows that profit patterns were relatively consistent over this period in a large number of countries. The results suggest that, in the majority of countries, the magnitude of profit ratios is not related to the different stages of the business cycle in each market but rather follows a medium-term trend Issuing business a) EU-25 comparison Of the 136 respondents reporting figures for the issuing of credit cards in 2004, only 20 issuer institutions reported a negative profit rate. The weighted profit ratio average is 65%. Graph 38 shows the distribution of profit ratios for all respondents in The distribution reveals that 68 out of 136 issuers reported profit ratios higher than the median value (61.4%) of the sample. Of this group, 34 issuers reported figures between 61.4% and 131.8%, while the remaining 34 issuers reported figures above 131.8%. As regards the issuers whose reporting figures were below the median value, it is interesting to note that 34 issuers reported profit ratios between -50% and 14.7% while the remaining 34 issuers reported figures between 14.7% and 61. 4%. 66

88 Graph 38 - Distribution of profit ratios in the issuing of credit cards, 2004 Density rate In the light of these results, it is clear that the credit card issuing business was very profitable in all 25 Member States in b) Country analysis As with the acquiring business, we have also carried out an analysis of the profitability of the issuing of credit cards at country level for Again, in order to avoid giving equal weight to small and large issuers in the determination of the overall profit ratio, the country profit ratio is a weighted average of all the issuers in the country in question (the weight is given by the total income of issuers). Graph 39 shows this weighted profit average ratio for all Member States for Graph 39 Profit ratio (%) Profit ratio differentials in credit card issuing, 2004 Weighted average=65% A B C D E F G H I J K L M N O P Q R S T U V W Y Z Member States Going by the results depicted in Graph 39, the income generated by the issuing of credit cards is higher than the associated costs in all 25 Member States. The weighted average profit ratios vary from 3% to 147%. The EU-25 weighted average is 65%, with 11 countries above this figure. As with acquiring, some of these profit ratios are based on only a limited number of observations, which means that results may not be entirely representative of profitability for a 67

89 given country. Additionally, we have no way of assessing in such cases whether the results are driven by some outliers in the sample. However, with the exception of two countries, the number of observations per country is clearly higher for issuing than for acquiring, which makes it interesting to assess the degree of variability of profit ratios within a country, i.e. whether profits in each country are evenly distributed among issuers. Taking as examples the four countries with a high number of observations and displaying a weighted average profit ratio above the EU-25 average, we can observe a fairly similar pattern as regards profit ratios: while almost all issuers reported positive profit ratios, there are some discrepancies in the profit ratios of some top issuing institutions. Looking in detail at the data, it is possible to conclude that the differences in cost structures may explain to a large extent the discrepancies in profit ratios among top issuing institutions. Therefore, the differences observed in the profit ratios of some top issuing institutions in the same country seem mainly to reflect a different level of efficiency and not a fierce competition on prices. As with the acquiring of credit cards, it is necessary to analyse further the dynamics of the profit ratios at country level over the period in order to detect the influence of different stages of the business cycle in each country. Graph 40 shows the country weighted profit ratios in the issuing of credit cards during the period Graph 40 profit ratio (%) Profit ratios in credit card issuing, A B C D E F G H I J K L M N O P Q R S T U V W Y Z Member States From the analysis of this graph, it is apparent that profit patterns were relatively consistent over this period in almost all countries 76. More importantly, these results suggest that, as for the acquiring business, the magnitude of profit ratios is not related to the different stages of the business cycle in each market but rather follows a medium-term trend. All in all, the issuing of credit cards seems to be highly profitable in the large majority of the EU-25 Member States. The persistence of high profit ratios over a relatively long period of time suggests that this might be the result of having and exerting market power Taken issuing and acquiring businesses together According to the mainstream theory on two-sided markets, the fact that benefits may arise jointly on the two sides of the industry means that there may be no meaningful economic relationship between benefits and costs on either side of the market considered by itself. Thus, it 76 The significant increase in the weighted profit ratio of country Q in 2004 is due to the inclusion of two important issuers in the sample only in this period. 68

90 is relevant to analyse the profitability patterns of both acquiring and issuing together, and to explore the findings in the light of this theory. Graph 41 shows the weighted average profit ratios of acquiring and issuing credit card businesses for all EU Member States for Graph Issuing Acquiring Issuing and Acquiring: Profit ratios, 2004 Profit rate (%) A B C D E F G H I J K L M N O P Q R S T U V W Y Z Member States An important conclusion that emerges from the analysis of this graph is that the issuing business is clearly more profitable than the acquiring business in almost all countries. In some countries the issuing business is 50 times more profitable than acquiring. How is it possible to explain the different profitability on issuing and acquiring for countries with both a relatively similar economic size and a similar level of interchange fee? In order to better understand the functioning of the credit card payment system as a whole, it is interesting to compare two similar institutions (1 and 2) in two different countries (G and O). These two institutions have a similar annual number of transactions in the acquiring and issuing business. However, one should bear in mind that the revenues and costs of these institutions for issuing and acquiring relate to all the payment systems in which they participate. These two institutions have very similar incomes from the acquiring business because, given that the MSC is the principal source of revenue, institutions 1 and 2 are charging a similar average merchant service charge per transaction. However, institution 2 incurs a higher cost for processing, which erodes its profit margin (the average interchange fee paid is relatively similar for both institutions). When one looks at the issuing side, the picture changes dramatically. As a matter of fact, while both institutions incur more or less the same costs and receive more or less the same average interchange fee per transaction, institution 2 receives much high cardholder fees. In conclusion, cost inefficiencies explain why institution 2 has lower profits than institution 1 in the acquiring business. Similarly, the level of cardholder fees explains why institution 2 is much more profitable than institution 1 on the issuing side. If one generalises this example to all institutions in both countries, the figures shown in Graph 41 are consistent with this. 69

91 2.1.4 Profitability vs interchange fee Of the 136 institutions reporting data for the credit card issuing business for 2004, 118 also reported data on the total revenue obtained through interchange fees. It should be noted that 100 of these 118 reported positive profit ratios. In order to quantify the importance of the interchange fee in the total income of the issuing institutions, we have carried out an additional exercise. We compared the total income of these 100 issuers, i.e. including the part generated by interchange fees, with the income that the same issuers would obtain if the interchange fees were taken out from their total revenue (which is equivalent to having a zero interchange fee). This exercise reveals that if that part of total income due to interchange fees were to be taken out, 62 of the 100 institutions reporting positive ratio profits would nevertheless remain profitable 77. These findings may partly be explained by the likelihood that the income from cardholder fees and interest may make issuing profitable anyway. Graph 42 shows the country weighted average profit ratio for the 118 issuers, both when that part of revenue due to interchange fees is included in total revenue and when it is excluded. Graph 42 Profit ratios with and without interchange fee, Profit rate with IF Profit rate without IF 100 Profit ratio (%) A B C D E F G H I J K L M N O P Q R S T U V W Y Z Member States The fact that a high number of issuing institutions remain profitable in the extreme situation of a zero interchange fee is relevant. From our exercise, it can be concluded that in 20 of the 25 countries, the interchange fee significantly adds to the positive level of profits in the credit card issuing business that would be obtained anyway with zero interchange fees. This exercise seems to partially invalidate one of the main results of the theoretical models described in the chapter on the economic literature, which suggest that a positive optimum level of interchange fee is needed because price market mechanisms fail to internalise the existing externalities, with the result that total system output would suffer if issuing were not subsidised through the transfer of revenues from acquirers. Naturally, the aim of this exercise is not to argue in favour of a zero interchange fee but, in the light of the results, it is legitimate to question the optimality of the current level of interchange fees in several countries. Our findings seem to confirm some recent theoretical predictions of the twosided market literature, which suggest that privately optimal interchange fees may be too high, notably if merchant fees increase with interchange fees but issuers do not pass the additional interchange fee revenue back to cardholders. In this case, high interchange fees are a way to 77 Naturally, it is straightforward to conclude that the number of profitable issuing institutions would be even greater for a reduced interchange income than for a zero income. 70

92 transfer profits to the side of the scheme where they are least likely to be competed away (Wright, op. cit. 2004). Similarly, these results also seem to cast substantial doubt on the justifications for the existence of interchange fees put forward by the payment card systems. For instance, one international network believes that in the absence of POS interchange fees paid by acquirers to issuers, issuers would have to recoup all of their costs from cardholders and this would lead to a level of card issuing that is not optimal for the system as a whole. This statement seems to be largely refuted by our results. The justification put forward by another international network, which considers that the interchange fee provides for a transfer of revenue between issuers and acquirers to achieve the optimal delivery of services by both acquirers and issuers to merchants and cardholders, is also not supported by our results. For instance, looking at country U in Graph 41, it can be seen that the issuing of credit cards is much more profitable than acquiring (which is even negative). Moreover, Graph 42 shows that issuing credit cards in country U would still be profitable even with a zero interchange fee. In such a context, the role of interchange fees as a mechanism to redress the imbalance between issuers and acquirers costs and revenues in delivering a payment card service is not readily understandable. 2.2 Debit cards Acquiring business a) EU-25 comparison Of the 53 respondents that reported figures for the acquiring of debit cards in 2004, 30 reported a positive and 21 a negative profit ratio. Graph 43 shows the distribution of profit ratios in the acquiring of debit cards for 2004 for all respondents in 19 countries 78. Graph 43 - Distribution of profit ratios in acquiring of debit cards, 2004 Density rateqd The distribution in Graph 43 reveals that reported profit ratios tend to be concentrated around zero (the median value is -3%). The overall weighted average profit ratio is 5% (weights are given by the total income of acquirers), which suggests that the acquiring of debit cards is on average significantly less profitable than the acquiring of credit cards at EU level. 78 Six countries were excluded from the analysis of debit card acquiring because of data unavailability. 71

93 b) Country analysis Turning now to the analysis of profitability at country level, Graph 44 shows the country weighted profit ratios in the acquiring of debit cards for Graph 44 profit ratio (%) Profit ratio differentials in debit card acquiring, 2004 A B C D E F G H I J K L M N O P Q R S Member States Weighted average=5% Graph 44 shows a strong variation on the weighted profit ratio across countries. The weighted country profit ratio varies from -32% to 35%. The acquiring of debit cards is only profitable in 12 out of 19 countries in the EU in As with the analysis of credit cards, some of these profit ratios are based on only a limited number of observations, which again means that results may not be entirely representative of profitability for a given country. In certain cases, the sole observations in the sample are representative of the country, given that they are from large and specialised acquiring institutions. However, some caution is necessary in cases where the country figure is based only on a small acquirer. For countries with a significant number of observations, we can assess the profit ratio variability within a country. In country D, for instance, we can conclude that acquiring is unprofitable because all 13 acquiring institutions reported negative profit ratios. In contrast, all 4 institutions in country S reported positive figures the two biggest institutions reporting profit ratios of around 50% which seems to prove that the acquiring of debit cards in country S is a profitable business. It is interesting to note that a similar pattern for these countries is found above for the acquiring of credit credits. Graph 45 reports the weighted profit average ratios by country over the period From the analysis of this graph, we can conclude that profit patterns remain consistent in at least a certain number of countries over this period of time. Accordingly, the magnitude of profit ratios in the acquiring of debit cards seems to follow a medium-term trend in these countries. 72

94 Graph 45 Profit ratio (%) Profit ratios in debit card acquiring, A B C D E F G H I J K L M N O P Q R S Member States For other countries, the results are somewhat more difficult to interpret. Some of these results are explained by the fact that some of the respondents are not represented in the sample over the entire period Issuing business a) EU-25 comparison Four countries were excluded from the analysis due to data unavailability. Of the 71 respondents reporting figures for the issuing of debit cards in 2004, 21 reported a negative profit ratio. Seventeen issuer institutions reported a profit ratio below -10%, 18 reported a profit ratio between -10% and 33%, 18 between 33% and 120% and 18 above 120%. Graph 46 shows the distribution of profit ratios in the issuing of debit cards for all respondents in Graph 46 - Distribution of profit ratios in the issuing of debit cards, 2004 Density rate If one also takes into consideration that the weighted profit ratio of all respondents for 2004 is 47% (weights are given by the total income of debit card issuers), it seems reasonable to conclude that the debit card issuing business is profitable at EU-21 level. As for credit cards, 73

95 these simple statistics show that issuing for debit cards is significantly more profitable than acquiring. Interestingly, the weighted profit average ratio for the issuing of debit cards in the EU-21 is lower than that for the issuing of credit cards in the EU-25 countries (65%). In order to draw a more meaningful conclusion about the level of profitability in these industries, we compared the profit ratios of issuers that issue both credit and debit cards. Using the figures reported by 54 issuers which issue both credit and debit cards, we obtain for 2004 a weighted profit ratio of 43% for debit cards and a weighted profit ratio of 63% for credit cards. This seems to prove that the issuing of credit cards is on average more profitable than the issuing of debit cards. b) Country analysis As with acquiring, we also carried out an analysis of the profitability of issuing at country level for Again, in order to avoid giving equal weight to small and big acquirers in the determination of the overall country profit ratio, the country profit ratio is a weighted average for all issuers of debit cards in the country in question (the weight is given by the total income of debit card issuers). Graph 47 shows this weighted profit average ratio for 21 Member States for Graph 47 Profit ratio (%) Profit ratio differentials in the debit card issuing business, 2004 A B C D E F G H I J K L M N O P Q R S T U Member States Weighted average = 47% As shown in Graph 47, the income generated by the issuing of credit cards is higher than the associated costs in 19 out of 21 Member States. The country weighted average profit ratios vary from -20% to more than 140%. It is interesting to note that all 6 respondents in country T reported positive rates. Looking at the two largest issuing institutions in this country, their profit ratios are 150% and 140%, respectively. This is clear evidence that issuing debit cards in country T is a very profitable activity. In some countries, issuing institutions are somewhat more heterogeneous. Looking in detail at the data, it is possible to conclude, again, that differences in cost structures may to a large extent explain the discrepancies in profit ratios among issuing institutions. Moreover, Graph 48 confirms that the pattern displayed by Graph 47 for 2004 is consistent over the period in a number of countries. 74

96 Graph 48 profit ratio (%) Profit ratios in debit cards issuing, A B C D E F G H I J K L M N O P Q R S T U Member States It seems unquestionable that for countries S and T, the existence and persistence of a very high profit ratio over a relatively long period of time seems to follow a medium-term trend. Unfortunately, it is not possible to draw any conclusion regarding country U because the sample does not contain observations for the entire period Taken issuing and acquiring businesses together An analysis of Graph 49, which gives a comparison of weighted average profit ratios for the issuing and acquiring of debit cards across countries, shows that, as for the credit card analysis, issuing is clearly more profitable than acquiring in almost all countries. 75

97 Graph 49 Profit ratio (%) Issuing and Acquiring: Profit ratios, 2004 Issuing Acquiring A B C D E F G H I J K L M N O P Q R Member States The reason why issuing is several times more profitable than acquiring in some countries, e.g. country S, is again the role played by the interchange and cardholder fees. In fact, while the level of the MSC charged by an acquirer in country S to merchants is very high, the amount of interchange fee per transaction, which is transferred to the issuing side, erodes its profit margin. This also confirms the results in the chapter on merchant service charges, where it is shown that this country had the highest weighted average interchange fee as a proportion of the weighted average merchant service charge in As regards the debit card issuing industry in country S, one can observe that the two largest issuers receive revenue from the interchange fee that is 5 times higher that the cost associated with the transactions carried out. Additionally, these issuers also receive a considerable amount in cardholder fees. Consequently, the issuing of debit cards is very profitable in this country. 3. Conclusion and analysis The issuing of credit cards is very profitable. On a pan-eu scale, credit card issuers had a weighted average profit-to-cost ratio of 65% in 2004 while debit card issuers had a weighted average profit ratio of 47%. In most EU Member States, the weighted average profit ratios remained fairly stable over the period 2000 to It therefore appears that in most countries the magnitude of profit ratios is not related to different stages of the business cycle in each market but rather follows a medium-term trend. Interchange fees appear to magnify these profits. It appears that 62% of all banks surveyed would still make profits with credit card issuing even if they did not receive any interchange fee revenues at all. In 23 EU Member States, at least one bank participating in the survey was able to make a profit from issuing credit cards without interchange fees. The weighted profit-to-cost ratio of all respondents for credit card acquiring was 15.9% in 2004, and for debit card acquirers 5%. In a large number of EU Member States, profit ratios remained fairly stable over the period 2000 to

98 The reason why issuing is several times more profitable than acquiring in the majority of countries is the role played by the interchange fee as a cost and revenue element in the payment card system. While the profit ratios of acquirers and issuers in some markets are homogeneous, banks in a few countries reported strongly diverging profit ratios. The individual balance sheets of acquiring and issuing banks showed that this is mainly due to the strongly varying cost structure of acquirers and issuers in these countries. Finally, it was found that two institutions with a similar size in two different countries where the level of interchange fee is similar may display different profitability ratios. Cost inefficiencies seem to explain the different profitability on acquiring while the level of cardholder fees seems to explain the different profitability on issuing. The analysis of profit ratios in POS payment card systems may provide valuable information for a competition analysis. In this respect, it is worth noting the following observations. First, the high and persistent profit ratios found by this inquiry in relatively mature markets, together with other evidence collected on entry barriers, suggest the existence and exercise of market power in these markets. Second, the question whether card issuers can offer payment cards at affordable prices to consumers in the absence of interchange revenues is of relevance for a competition analysis of interchange fee agreements. If the transfer of revenues were necessary for the operation of a payment card system, then interchange fee agreements may not be caught by Article 81(1) EC, even if the fees determine the prices charged by an acquirer to merchants. However, the above findings on the profitability of payment card issuing cast doubt on the assumption that in the absence of interchange fees, issuers could not recoup their costs from cardholders. This observation does not exclude, however, that the use of interchange fees may lead to the more efficient operation of a POS system. However, it does seems to confirm some recent theoretical predictions in the literature on two-sided markets suggesting that privately optimal interchange fees may be too high, notably if merchant fees increase with interchange fees but issuers do not pass the additional interchange fee revenue back to cardholders. In this case, high interchange fees are a way to transfer rents to the side of the scheme where they are least likely to be competed away. Finally, the cost structure of acquirers may provide valuable information on the reasons for high market concentrations. If fixed costs rather than variable costs were the reason for the low profit margins of banks in a given country, the need for scale may be the principal reason why there are only a few acquirers in this market. In this respect, while fixed costs may be of importance for some individual banks, it would appear in general that the costs of acquirers are determined by interchange fees, which are variable costs. This cost structure for most of the acquirers surveyed raises the question whether reasons other than scale may be responsible for the high concentration of the acquiring business in some countries, such as the existence of interchange fees. 77

99 Section C Organization of the Industry

100 VIII. Concentration of Acquiring and Issuing Businesses This chapter aims to analyse the degree of concentration on the acquiring and issuing sides of the payment card markets across the EU-25. It is interesting to see whether acquiring is really as competitive as some of the literature claims it to be, and if not, whether there are any particular reasons for the level of concentration in certain Member States. 1. Acquiring The concentration of acquiring activity is assessed in terms of how many institutions offer payment card acquiring services in a given Member State. The degree of concentration is assessed separately for each payment card network. The analysis, therefore, looks separately into international credit, international debit and national networks. It needs to be noted, however, that a concentration analysis says nothing by itself about possible coordination in the market. Moreover, it says nothing about whether a monopoly acquirer, should one be identified, extracts significant rents through exercising its market power. What this analysis also does not assess is a set of other nonetheless important factors, such as, for example, whether acquiring is vertically integrated, whether all acquirers jointly own a processor, or whether the scheme (i.e. set of rules ) is managed by an association of acquiring banks, even though each bank has an independent acquiring licence. All these conditions may affect competition amongst market players. Furthermore, some merchants, particularly larger ones, that choose to accept international payment cards (among others) are just as likely to be acquired by cross-border acquirers. Under these conditions, cross-border acquirers may join in for the competition for these merchants with domestic acquirers. The analysis, nonetheless, does not look at overall concentration in acquiring but rather limits itself to domestic concentration (i.e. excluding the aspect of cross-border acquiring). 1.1 Acquiring in international payment networks: credit cards An analysis of the level of concentration yields some interesting results. Graph 50 shows the level of concentration in the acquiring of credit cards for one of the international networks, measured in terms of the Herfindahl-Hirschman Index (HHI), as well as the number of acquirers across the EU-25 Member States for An HHI of up to 2000 is assumed to raise no substantial competition concerns. The graphical analysis shows that acquiring is quite concentrated across the EU-25. Spain is the only country where the level of concentration is below In five Member States, namely Austria, Cyprus, Denmark, Finland and Portugal, acquiring for MasterCard credit card transactions is performed by a single institution. Furthermore, three others (Luxembourg, Netherlands, Germany 79 ) have an HHI higher than 8000, meaning that acquiring is very concentrated, albeit not in the hands of a sole acquirer. In some countries, one player may have around 90% of the total acquiring volume for the MasterCard credit card transactions of the country. The average HHI across the EU-25 is about The number of acquiring institutions per Member State allows us to assess, in addition to the concentration index, whether a high concentration signalled by high HHI is caused by a low 79 These numbers rely solely on data provided by the network. The network reported the aggregate acquiring turnover under the name of a licensing company. This aggregate turnover may in fact represent the cumulative acquiring turnover of at least 3 network licensees. The individual turnover data of these licensees were not supplied by the network and therefore are not considered in the analysis. Cross-border acquirers are not accounted for either. For these reasons, the level of concentration shown in the graph 50 may in fact significantly overestimate the true one. For the sake of consistency, however, it was decided to present the network s reported numbers as is (with no adjustments). 79

101 number of acquirers as such, or rather by big market shares held by a few/single players in the market. As the graph shows, apart from a few spikes, a high concentration goes hand in hand with a small number of acquirers in a particular Member State. Spain, for example, has a low concentration but also the highest number of banks (19) performing acquiring in the network considered, which is in line with this first explanation. Spikes are also seen, on the other hand, in countries where the level of concentration remains quite high despite a large number of acquirers. Examples include Slovakia, the UK, and Hungary. However, this can be explained by the second argument that a large number of banks can still result in a high concentration where market share is unequally distributed among acquirers. Graph 50 HHI Index Level of Concentration (HHI Index and Number of Acquirers Across EU) in an International Network (Credit Cards), 2004 HHI No of Acquirers Number of Acquirers 0 ES LV IT CZ SK SE GR PL UK IE LT ET HU AVE MT BE LU NL DE* AT CY DK FI PT * - DE data need to be treated with caution (see footnote in the text) Member States 0 A simplified analysis of the dynamics in the level of the HHI across the EU-25 (see graph 51) reveals that across the EU-25 there are countries with both falling and rising levels of concentration over the period from 2000 to Rather than focusing on the direction of change, the analysis starts with an assessment of the absolute changes. For the sake of simplification, countries with an absolute change of 10% or less are considered to be countries with insignificant variation in the level of concentration and are therefore are not analysed in detail. In contrast, countries with absolute variations exceeding the 10% threshold are scrutinised further. Thus, it turns out that the majority of the countries with significant changes in the degree of concentration over the period examined are new Member States, i.e. the Czech Republic, Hungary, Latvia, Lithuania, Malta, Poland and Slovakia. Since these countries are characterised by quite immature and unstable payment card markets, these changes may be explained by market adjustments. No consistent pattern of change is noticeable across these countries. Among countries with significant variation in the level of concentration, the following three old Member States stand out: Greece, Ireland and Luxembourg. A detailed analysis of the cause of such significant changes in the acquiring concentration has been performed in order to understand the underlying reasons in each of these geographical markets. In Greece, a 25% decrease in the HHI can mostly be explained by the fact that, as of 2000, more acquirers joined the acquiring market for the credit cards of the network considered, therefore raising the total number of acquirers. Simultaneously, the share of the biggest player fell over the same period, further pushing down the concentration index. In Ireland, however, a 25% drop in the HHI was due to slightly different reasons. Even though the overall number of acquirers went down, the downward pressure on the index was caused by a significant reshuffling of the market shares of the three biggest banks. The biggest acquirer in 2000 had lost by 2004 a significant part of its market share to its competitors. This 80

102 brought down the concentration, as the strong market position of the main acquirer had been weakened by the increased shares of the other two competitors. In Luxembourg, on the other hand, the concentration index increased by an astonishing 40%. In fact, this was the highest relative growth in the concentration index across the whole EU-25 (ahead of Slovakia with 33%). A detailed analysis suggests that the increased concentration was mostly due to the fact that from 2000 the main acquirer had been constantly expanding its market share for credit cards, while the share of its main competitor had been gradually falling. Graph Change, % 60% 40% 20% 0% -20% -40% -60% Change in HHI Index from 2000 to 2004 in the International Credit Card Network across EU Member States, % -52% -44% -38% -25% -25% -22% 2% 2% 2% 7% 10% 16% 31% 33% 40% CZ PL LV GR IE LT NL DE * UK SE ES ET IT MT HU SK LU -7% -5% Member States * - Dynamics in DE HHI should be treated with caution giv en the data qualif ication provided in f ootnote in the text. In f act, there was an increased de-concentration in German acquiring market f or credit cards in the analysed network ov er the examined period, howev er, presented results rely solely on data supplied by the network. 1.2 Acquiring in international payment networks: debit cards According to data supplied by one of the international networks, the total merchant sales volume, i.e. acquiring volume, on debit cards in the EU-25 had increased by more than 400% by In 2001, the UK alone accounted for more than 70% of this volume. By 2004, however, its relative share had fallen substantially to about 52% (30% drop) (see Graph 52). In 2001, more than 85% of the total acquiring turnover on debit cards in the network in question was generated just in the UK and Spain. By 2004, however, the share of Spain had fallen (by about 20%), while that of other Member States had increased significantly. Thus, the share of acquiring volume generated in Italy grew from a moderate 4% in 2001 to 21% in 2004, corresponding to a relative growth of more than 450%. Similarly, the relative share of the Austrian acquiring volume on debit cards in the network grew from 1.5% to 12%, amounting to almost 700% growth in relative terms. As a result, the combined share of the UK, Italy and Austria in 2004 accounted for about 85% of the total acquiring volume for the debit cards of the network in the EU

103 Graph Versus 2004: Changes in Sharesof MemberStates in Overall EU Acquiring for Debit Cards in the International Network, %* UK 51,7% 73,2% ES 12,0% 9,2% 2001 IT 3,9% 21,3% 2004 PL 3,5% 1,4% Member States PT AT HU 1,5% 0,1% 1,5% 0,9% 1% 11,8% SE 0,8% 1,4% LU 0,8% 0,3% CZ 0,7% 0,6% SK 0,5% 0,3% 0,0% 10,0% 20,0% 30,0% 40,0% 50,0% 60,0% 70,0% * Data source: The network's data MC Acquiring Share, % The analysis of the level of concentration across the EU-25 in the acquiring of debit cards in the network considered (see Graph 53) suggests that, as with the acquiring of credit cards, the acquiring of debit cards is characterised by a high degree of concentration. Only Spain (HHI of 1464) and Latvia (HHI of 2018) are below or slightly above an HHI of The same group of countries as for credit cards (with the exception of Luxembourg and Finland) report a single acquirer for debit cards, namely Cyprus, Denmark, Luxembourg and Portugal. Graph HHI Index Level of Concentration (HHI Index and Number of Acquirers Across EU) in the International Network (Debit Cards), 2004* HHI No of Acquirers Number of Acquirers 0 ES LV SK GR PL CZ UK LT MT ET AVE SE IT HU AT CY DK LU PT * - Based only on the netw ork's reported data Member States 0 80 This graph does not include acquiring volumes on co-branded cards if another facility (not that of the network considered) was used for transactions. Thus, no volumes are reported for some countries. 82

104 The analysis of the dynamics in the HHI shows that 5 of the 18 countries analysed, which also happen to be countries with a single acquirer, show no changes in terms of their respective levels of concentration over the period (i.e. Austria, Cyprus, Denmark, Luxembourg and Portugal) (see Graph 54). As previously, Member States with a significant variation (more than 10%) were analysed further. Thus, as with the acquiring of credit cards, most of the countries with a substantial variation over the period examined were new Member States (Estonia, Hungary, Latvia, Lithuania, Poland) with quite immature acquiring markets and therefore possibly unstable acquiring. As before, no particular pattern can be identified (both increasing and decreasing concentrations are equally reported). Among the old Member States, a substantial variation in concentration levels was observed in four countries: Greece, Spain, Sweden and the UK. In Greece, despite a fall in the number of acquirers for the network s debit transactions, the HHI concentration index went down by almost 60%. That was mostly due to the fact that the biggest acquirer in 2001, which acquired most transactions back then, subsequently lost a significant portion of its market share to its competitors in the following years. In fact, this acquirer was no longer the market share leader by Therefore, the concentration index was brought down mostly through the redistribution of acquiring market shares among the existing acquiring institutions. A similar situation was observed in Spain, where the HHI also dropped by almost 60%. Here, in 2001 two acquirers jointly owned more than 80% of the total acquiring market for international debit card transactions in the network. By 2004, however, their joint share had dropped significantly, with other acquirers gaining an increasingly higher share of the acquiring market. In the UK, a similar drop of almost 60% in the concentration index was caused primarily by two things. First, the overall number of acquirers for debit card transactions in the network increased over the period. At the same time, the market share leader of 2001 had lost a substantial part of its market by On the other hand, two other acquirers had very much strengthened their position. The level of concentration had actually grown in Sweden by This was mainly due to the fact that the total number of acquirers fell over the period examined. Furthermore, the market leader had expanded its market share in the acquiring of debit card transactions in the network considered. Graph Change, % 60% 40% 20% 0% -20% -40% -60% -80% Change in HHI Index from 2001 to 2004 in the International Debit Card Network across EU Member States, % GR UK ES LT LV PL SK IT MT CZ ET SE HU -11% -6% -4% -1% -25% -59% -58% -57% -40% Member States 10% 15% 26% 37% 83

105 1.3 Acquiring in national payment networks A similar analysis of the level of concentration has been performed for national networks. Only Member States with operating national networks with available transaction volume data have been considered. The analysis does not cover, among others, such national networks as Aurore (Cetelem, France), Moneta (Setefi, Italy), and Activa (KoperBanka, Slovenia). Nor does it look into Spanish networks, such as ServiRed, Sistema 4B and Euro 6000, because most of the cards issued in these networks are co-branded with international payment networks and are therefore analysed above. Finally, many of the domestic debit cards issued in the national networks tend to be cobranded with international network brands, such as Maestro and Visa debit (in rare cases, for example in Finland, the domestic debit card Pankkikortti is co-branded with an international credit card logo such as MasterCard and Visa). Therefore, some of the volume reported for the national network may actually include turnover data reported by international card networks (such as MasterCard and Visa), leading to a possible double-counting of the market size as well as to the potential overstatement of the relative shares of each individual national network. This parameter has not been controlled for, and may therefore lead to a possible bias in the results. The analysis looked into ten national (domestic) networks (see Graph 55). The joint acquiring turnover on these networks increased by about 50% in relative terms over the period considered. This cumulative turnover was more than four times higher than the turnover generated on one of the international debit networks in the same period. This result would be even more striking if the turnover generated in the Spanish national networks (ServiRed, Sistema 4B and Euro 6000) had been included. Graph 55 Shares of Member States in the Total Acquiring Volume for Domestic Cards, 2004, %* Network 1 45% Network 2 14% Network 3 11% Network 4 10% National Networks Network 5 Network 6 Network 7 4% 6% 7% Network 8 2% Network 9 Network 10 0% 1% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% * Data source: Data from National Networks MC Acquiring Share, % As the next graph shows (Graph 56), most national networks are characterised by a very high level of concentration. In fact, 4 of the 10 networks analysed, according to the reported data, had a single acquirer in 2004 (Belgian Bancontact, Danish Dankort, Maltese Cashlink and Dutch Currence (PIN)). It needs to be noted, however, that following an investigation by the Dutch competition authority, Interpay (the operator of Currence (PIN) network) had to begin gradually transferring merchant contracts to individual Dutch banks in 2004, thus meaning there 84

106 was no longer a single acquirer for domestic cards. In Germany, competition exists at the level of network service provider, who assumes many of the functions performed by acquirers in other networks. Consequently, a single de jure acquirer in Germany (ZKA) does not lead to a de facto concentration in acquiring 81. Hungarian GiroBankCard also reported a high level of concentration (HHI of more than 9000), owing to the fact that the bulk of acquiring turnover is handled by a single bank, K&H Bank. The lowest level of concentration is found in the Italian national network Cogeban (Pagobancomat), with an HHI below 600. This low HHI was primarily due to an extremely high number of acquirers in the market (about 430) as well as the fact that no single acquirer has a large share of the market. The second lowest level of concentration is observed in the French national network Groupement de Cartes Bancaires, with an HHI slightly above In 2004, there were about 80 active members in the network, though only four, generated over 50% of the overall acquiring turnover of CB. HHI Index Pagobancomat 430 (!) (IT) Acquirers Graph 56 Level of Concentration (HHI Index and Number of Acquirers Across EU) in Domestic Networks, 2004* Carte Bleue (FR) Pankkikortti (FI) Laser Card (IE) Adjusted AVE (no Pagobancomat) GiroBankCard (HU) * Data source: Data from Domestic (National) Debit Networks ** This is state of playuntil March 2004, when Interpay (BeaNet) started the transfer of merchant contracts to banks Bancontact Dankort (DK) Cashlink (MT) PIN (NL)** (BE) National Networks HHI No of Acquirers Number of Acquirers Given the low level of concentration in Italy, its HHI has not been taken into account in calculating the average level of concentration and average number of acquirers in national networks across the EU-25. Were it to be included, this could create a significant bias in the overall result and therefore lead to misjudgment of the average level of concentration across the whole EU-25. Thus, excluding Italy, the average level of concentration in the EU-25 in 2004 was about 7100, which exceeded by about 24% the corresponding average level of concentration in the international networks considered (both debit and credit). The average number of acquirers, again adjusted to exclude the Italian Pagobancomat, was 13. When also adjusted for the relatively high number of French acquirers (80), the average figure fell to mere 3 acquirers per national network. This result means that concentration in national networks remains very high across Member States. Nonetheless, as pointed out earlier in this section, this analysis alone is sufficient to determine whether single acquirers in these markets abuse their market power in an attempt to extract excessive rents, among other things to provide issuers with substantial gains through the interchange fee mechanism. What needs to be analysed further is whether a high level of domestic concentration actually leads to higher prices, i.e. for the MSC, that is to say, 81 Due to the peculiarities of the German acquiring market, it is not included in Graph 56, as it could have given the misleading impression of an overly concentrated market (HHI of due to single de jure acquirer). 85

107 whether the level of concentration in a given country positively correlates with the level of the MSC charged in that country. 1.4 Analysis of the correlation between the HHI level, the number of acquirers and the MSC level An analysis has been performed to determine the correlation between the level of the concentration index (HHI) and that of the weighted average MSC as well as between the number of acquirers in a given country and the MSC level. Table 6 summarises the results of the correlation analysis. The analysis did not lead to any conclusive results. Even though the correlation between the HHI level and the level of the weighted average MSC in both international and national networks seemed to be positive (meaning that higher concentrations on average lead to higher levels of MSC), the absolute level of the correlation coefficient did not exceed 50% 82, with the highest correlation coefficient being recorded in national networks. Nevertheless, it needs to be borne in mind that this correlation analysis does not control for other variables that may affect the MSC level. These other variables may include interventions of national competition authorities, the average cost of acquiring in the country, etc. Table 6 Degree of Correlation between level of HHI, number of acquirers and level of MSC across the EU-25, 2004, % International Credit Network International Debit Network National Network Corrrelation b/w HHI index and Level of W/A MSC 22,3% -0,5% 41,8% Corrrelation b/w No of Acquirers and Level of W/A MSC -9,4% -17,4% 2,9% 2. Issuing The analysis of the level of concentration on the issuing side of the payment card market yielded no evidence of excessive concentration across the EU-25 Member States. Generally, issuing is characterised by a high number of market players of varying size. No cases of a single issuer have been reported. 3. Conclusion and analysis The economic literature seems to claim that the structure of two-sided markets leads to an outcome where acquiring is competitive and issuing is not. Evidence on market structures across the EU-25, however, seems to contradict this view. Acquiring, whether due to historical or other reasons, seems to be quite concentrated across many EU-25 Member States. Issuing, on the other hand, is much less concentrated. The business of acquiring credit cards in the international networks appears highly concentrated. Looking at one of the international systems, where acquiring for credit cards has nearly doubled in the period from 2000 to 2004, it appears that the EU average concentration index HHI is quite high, reaching 5800 in This is almost three times the threshold of 2000 above which competition concerns start to arise. In this international system, the acquiring of credit card transactions is performed by a single institution in five Member States, while in three other Member States the HHI is higher than 8000, with one institution often having more than 90% of the total acquiring volume. A dynamic analysis of this market concentration in the 82 The absence of a correlation does not prove the statistical independence of two sets of numbers, but a correlation coefficient approaching unity does indicate a strong statistical dependence. 86

108 period from 2000 to 2004 shows that in 10 EU Member States, the variation in the HHI was above 10%. Seven of these countries are new Member States. The business of acquiring debit cards in the international systems appears equally highly concentrated. Looking at one of the international systems, where this business expanded by more than 400% between 2000 and 2004, the average HHI exceeded the threshold of 2000 in 16 Member States. Five of the 18 countries analysed in the time series (2000 to 2004) showed no change in the concentration index at all. As with credit cards, the Member States with significant changes (above 10%) in the concentration index were mainly new Member States. Turning to national card payment systems, ten systems were analysed. The joint acquiring volume of these national systems grew by 50% between 2000 and The majority of national networks are characterised by a very high level of concentration in the acquiring business. The adjusted average level of concentration in these ten systems was about 7100 in 2004, which exceeds the corresponding level in the one international system considered by 24%. A correlation analysis comparing HHI levels and the average levels of merchant service charges (MSCs) does not lead to conclusive results. Even though the correlation is positive for both the national systems and the one international system considered, the absolute level of the correlation coefficient does not exceed 50%. However, this correlation analysis does not control for other variables which may affect the level of the merchant rates. 87

109 IX. Integration of Card Payment Systems 1. Different degrees of vertical integration of card payment systems in the EU In the EU-25, a wide range of different card payment systems with a varying degree of vertical integration can be observed. According to the degree of vertical integration, the various services described above may either be reserved to one or a few entities or be subject to competition among banks and non-bank institutions. The Commission has analysed the most important POS card payment systems within the EU. The industry generally distinguishes between open or four-party card payment systems and closed or three-party card payment systems, where the scheme owner also engages in the financial aspects of the payment card business by issuing cards and acquiring merchants. This is the case for American Express, Citibank (Diners Club) and JCB, which (mainly) issue and acquire cards themselves. These systems are also referred to as proprietary systems, as the scheme owner typically is the proprietor of the technical network used for routing, switching, clearing and processing the transactions. The industry also calls them T&E card systems as these systems predominantly target cardholders who use cards in the travel and entertainment (T&E) industry. This typological distinction between open and closed systems may, however, be too rigid to capture the full complexity of the payment cards business in the EU-25, as the boundaries between typical open and typical closed card payment systems are fluid and may change over time. So-called open systems, for instance, have a varying degree of openness, as a scheme owner may decide to outsource issuing to individual banks but keep merchant acquiring in-house. Likewise, formerly closed systems such as American Express and JCB may wish to expand their activity by entering into joint ventures with local banks or by licensing banks to issue their cards. For a competition analysis, it may therefore be more appropriate to categorise POS card payment systems by their varying degrees of vertical integration as opposed to adopting a rather rigid classification into closed and open systems. The categorisation used here distinguishes those features of a card payment system that could be opened up to competition. We propose the following categorisation. Systems where the entity owning the card brand essentially does not engage in any activity other than setting the parameters for access to the network and the technical standards operate at level 1. Here, scheme ownership is legally separated from network ownership and the financial business of issuing and acquiring. Where a scheme owner engages in further technical or financial parts of the cards business, further integration levels are reached as follows: + 1 level : scheme owner switches authorisation requests itself + 1 level: scheme owner authorises and processes transactions + 1 level : scheme owner clears and/or settles transactions + 1 level: scheme owner acquires merchants + 1 level : scheme owner sells and/or rents POS equipment Thus, the minimum integration level is 1 while the maximum is 6. Within this framework, the most important domestic POS card payment systems in the EU, i.e.: systems that operate exclusively on a domestic basis, could be classified as follows: 88

110 Table IE NL DK FI - DE FR ES ES ES PT BE It should be noted that in the Netherlands and Denmark the separation of scheme ownership and network operator is only formal as most of the banks that co-own the scheme also appear to co-own the network operator. Taking this into account, it may be more appropriate to assign an integration level of 4 to these schemes. The international card payment systems MasterCard, Visa, Diners Club, American Express and JCB are more complex. Broadly speaking they could be classified as follows: Table MC VISA JCB Amex Diners MasterCard and Visa are not only the scheme owners, they also clear and settle transactions (virtually all cross-border and some domestic payment card transactions), and they sometimes also authorise transactions if an issuing bank cannot be reached for technical reasons (so-called stand-in authorisations). Moreover, in some EU Member States MasterCard and Visa also process transactions on behalf of local member banks. Thus, the precise level of integration of both systems may vary from one EU Member State to another. Amex, Diners and JCB are vertically integrated to a higher degree than Visa and MasterCard as they typically do issuing and acquiring themselves. The above tables provide a rough illustration of the significant structural differences between card payment systems and the possible scope for more competition within them. While domestic systems such as those in Ireland, the Netherlands, Denmark, Finland, France and Germany have legally separated scheme ownership from the technical and financial aspects of the payment cards business, other systems such as those in Belgium, Spain and Portugal have not yet taken this step and remain vertically integrated to a higher degree. This comparison is only the first step in a complex analysis and no quick conclusions should be drawn from the above tables. Even systems with an integration degree of 1 may be difficult to penetrate for foreign banks (and non-bank service providers), because the separation of the card scheme from the technical/financial aspects of the payment cards business alone may not be sufficient to allow for real competition. This may be the case because other barriers to competition exist or because a scheme s principal members may be able to adopt measures to reduce or eliminate the scope for competition. 89

111 Despite these caveats, the separation of scheme ownership, network operation and the financial aspects of the payment cards business, i.e. issuing and acquiring, may be a first important step towards more competition within a POS card payment system. Case study: Partial de-integration of the MasterCard system in Austria The evolution from a fully vertically integrated payment card system to a more competitive system may occur gradually, as exemplified by the MasterCard system in Austria. Here, the local banks set up an inter-bank company to create a domestic debit card system in 1980, which migrated to the Maestro platform in The banks decided to split the network operation from the financial aspects of the business. While their joint venture, Europay Austria, was to continue acquiring merchants, its processing activities were outsourced to another 100% subsidiary company of the banks, APSS. In November 2005, the banks finally relinquished all their shares in APSS. It is reported that the banks have entirely sold the processing business to an independent international card processor, First Data. Such sale of the processing business to an independent party turns the formal separation of the financial and the technical aspects into a real one. It provides a basis for competition between card processors in Austria, as APSS is now an independent processor who could compete with other international processors such as the Italian/Belgian/Dutch processor SinSys. The separation of processing and acquiring appears to be an important first step from a competition perspective, but Europay Austria still acquires virtually all domestic MasterCard and Maestro transactions, sells (and thereby controls) POS terminals to merchants and also remains the owner of specifications for communication standards. The extent to which a card system is vertically integrated matters from a competition viewpoint, particularly with respect to the financial aspects of the cards business. These relate to the guarantees given by banks to both cardholders and merchants that a transaction will be settled if all formal requirements 83 are fulfilled. For issuing banks, the risk calls for, amongst other things, the careful assessment of a cardholder s creditworthiness and for acquiring banks it entails, amongst other things, the evaluation of fraud risks at merchant outlets 84. Where interbank associations issue cards and/or acquire merchants on behalf of shareholder banks, the price for these financial services is not subject to competition between these banks. Examples include Banksys in Belgium and many inter-bank associations for acquiring merchants in the MasterCard and Visa systems (see the next section). 2. Joint ventures for acquiring services A closer look at the international card payment systems MasterCard and Visa shows that in many EU countries member banks of these open systems concentrate financial acquiring in the hands of a joint venture. Merchants may therefore face one single offer instead of many offers from competing banks. The table below shows the eight EU countries where MasterCard and/or Visa member banks have set up joint ventures for acquiring merchants. 83 For example, the collection of a signature on a receipt for the acceptance of a POS transaction. 84 As well as the risk of incurring chargeback losses if a merchant goes bankrupt after a transaction is contested by a cardholder. 90

112 Table 9 Acquiring done by interbank associations National system Austria X X Belgium X X X Denmark X X X Germany X X Finland X X Italy X X The Netherlands X X (until 2004) Portugal X X In Austria, Maestro debit card transactions and MasterCard credit card transactions are acquired by a joint venture of Austrian commercial and savings banks called Europay Austria. Likewise, Visa s member banks in Austria concentrate credit card acquiring in the hands of a joint venture, Visa Austria. In Belgium, member banks of the Bancontact/Mr. Cash system have set up Banksys to run the network and to acquire merchants. Banksys is the only acquirer for Bancontact/Mr Cash debit cards in Belgium and is co-owned by Belgian banks. It is the result of a merger between the Bancontact and Mr. Cash systems in Belgian banks moreover concentrate acquiring for MasterCard and Visa transactions in the hands of another joint venture, Bank Card Company (BCC). This joint venture has been acquiring Visa transactions since 1982 and MasterCard transactions since 1993 (merger with Eurocard Belgium). BCC is the largest acquirer for MasterCard and Visa in Belgium. In Denmark, the joint venture PBS acquires merchants for all international payment cards, i.e. Maestro, MasterCard, Visa and Visa electron, as well as JCB. PBS is co-owned by Danish banks and its acquiring activity dates back to the 1970s. In Germany, the inter-bank associations Concardis, B + S Card Service GmbH and Card Process GmbH acquire merchant contracts for MasterCard and Visa. In Finland, banks have concentrated merchant acquiring for international payment cards in the hands of Luottokunta. Luottokunta is co-owned by some Finnish banks and, to a smaller extent, by individual merchants. Luottokunta acquires the Maestro, MasterCard, Visa, Visa electron and OK cards. Its acquiring activity for the Visa brand dates back to 1980, for MasterCard to In Italy, banks have set up a joint venture called CartaSi for acquiring merchants. Its acquiring activity for MasterCard dates back to 1985 and for Visa to Until 2003, a committee of CartaSi recommended merchant fees and trading conditions, which the banks had to apply in the absence of explicit consent from CartaSi to do otherwise. The activity of the committee, however, had to be discontinued after a ruling of the Italian central bank, which held that the activity of the committee was illegal under Italian law. Today, merchant fees are determined by the individual CartaSi member banks that have a banking relationship with the merchant. In the Netherlands, PaySquare BV is a joint venture of Dutch banks for acquiring merchants for MasterCard and Visa credit card acceptance. PaySquare BV was founded in 2004 through a legal separation from Interpay Nederland BV. However, PaySquare remains a 100% subsidiary of Interpay, which in turn is fully owned by Dutch banks. Paysquare s predecessor Interpay started acquiring MasterCard/Eurocard transactions in 1980 and Visa transactions in